INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 90
Advancements in Oncological Dynamic Contrast-Enhanced Mri: A
Review and Critical Analysis of Prior Studies
Faizan Farooq
1
, Mohit Sharma
1
, Khursheed Ahmad
1
, Srishti Bhardwaj
1
and Arti
2
1
Department of Allied Health Sciences, Chitkara School of Health Sciences, Chitkara University,
Punjab
2
Department of Life Sciences Rayat Bahra University, Punjab-140103
DOI :
https://doi.org/10.51583/IJLTEMAS.2025.1412000010
Received: 13 December 2025; Accepted: 19 December 2025; Published: 26 December 2025
ABSTRACT
Dynamic contrast-enhanced magnetic-resonance imaging(DCE-MRI) has emerged as a transformative method
in oncological imaging, contributing to findings on tumor vascularity, permeability, and treatment response. This
review outlines an extensive assessment of DCE-MRI implementation across various malignancies, with a
specific emphasis on liver tumors, spinal metastases, and breast cancer. Through examination key
pharmacokinetic parameter such as Ktrans, Ve and Kep, the article highlights their diagnostic and prognostic
value in discriminating lesion types and forecasting microvascular invasion. It also delves into the integration of
artificial intelligence and radiomics in optimizing the interpretability and reproducibility of DCE-MRI data. The
review confronts issues related to motion artifacts, regulation, and crossplatform variability, while
recommending future directions or clinical uptake and research evolution. Through this synthesis, DCE-MRI is
placed as a crucial tool in precision oncology, with the extension of implementation across anatomical systems.
Key Words: Dynamic contrast-enhanced magnetic contrast resonance imaging (DCE-MRI), forward volume
transfer (Ktrans), reverse constant (Kep), Computed tomography (CT) and extravascular extracellular space
volume fraction ( VeVe)
INTRODUCTION
An operational imaging approach, Dynamic Contrast-Enhanced Magnetic Resonance imaging (DCE-MRI) that
facilitates neumarical evaluation of blood flow, tissue perfusion, and vascularity by examining signal refinement
curves subsequent to contrast administration [1]. This approach records dynamic shifts in tumor vascularity
permeability, often influenced by antiangiogenic therapies, delivering pivotal revelations into tumor biology and
response to treatment [2]. DCE-MRI creates Pharmacokinetic variables such as the forward volume transfer
(Ktrans), reverse constant (Kep), and extravascular extracellular space volume fraction (Ve), each indicating
distinct elements of the tumor microenvironment and vascular behavior. These parameters facilitate monitoring
of therapeutic effectiveness [3]. DCE-MRI proposes superior sensitivity than conventional MRI in recognizing
subtle changes within the tumor microenvironment, leading to a significant early biomarker of treatment
response [4]. This approach assists in voxel-wise calculations of perfusion characteristics, facilitating exact
outlining of tumor heterogeneity and permitting personalized treatment strategies. Clinically-MRI has shown
utility in multiple cancer: in hepatocellular carcinoma (HCC), it assists in assessing microvascular invasion
(MVI) and distinguishing HCC from hepatic adenomas; in cervical cancers, it elevates diagnostic confidence
and help avoid unnecessary radical surgeries [5]. and in soft tissue tumors, combined multiparametric analysis
with diffusion imaging have diagnostic accuracy [6]. Furthermore, quantitative parameters such as Ktrans, Kvp,
Ve, and add to tumor sorting and upgrade early prediction of treatment failure and recurrence risk [7]. Elevated
Ktrans values are often associated with increased vascular permeability characteristic of malignant tumors, while
higher Kep can reflect aggressive tumor behavior and enhanced perfusion [8,9]. Increased Ve typically indicates
tumors with extensive extracellular matrix or edema, suggesting potential for aggressive spread [5]. These
capabilities, such as planning and adaptive radiotherapy, enhance precision in tumor targeting [10].
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 91
Ultrasound has a low sensitivity for early (HCC), generally varying from 60% to 70% which can significantly
obstruct identification. Additionally, the results could be fluctuating due to operator-dependent factors, resulting
in unpredictable findings [9]. Computed tomography (CT) demonstrates balanced specificity for HCC detection,
generally between 70% and 80%. This limitation is aggravated by the risk related to ionizing radiation effects,
particularly troubling patients who require frequent imaging due to high risk of reappearance [8].
Standard MRI is vulnerable to motion distortion, notably those done by patient breathing, which can compromise
image quality. Furthermore, its precision declines when assessing small lesions, making it less effective in
identifying early-stage HCC. DCE-MRI offers numerical analysis, such as Ktrans which reflect the measure of
contrast agent transfer out of blood vessels within the extracellular medium. This parameter is vital to distinguish
benign and malignant tumors which generally exhibit higher Ktrans values because of elevated vascular
permeability and blood dynamics [9].
DCE-MRI offers quantitative evaluations of vascular characteristics, which are crucial for understanding tumor
microenvironments and hemodynamics, thus elevating tumor characterization compared to standard MRI
techniques [11]. The technique’s ability to assess tumor vascularity and permeability provides insights into tumor
behavior, which is key to therapeutic planning and enhancing patients result [12]. DCE-MRI can also facilitate
the differentiation of malignant and benign tumors by sharing elaborated findings of tumor blood flow, which is
often challenging with conventional imaging [13]. DCE-MRI can detect changes in tumor vascularity and
microenvironment, providing critical insights that reduce the need for invasive biopsies in assessing response to
therapy [14].
This investigation seeks to examine the reliability of (DCE-MRI) in cetegorizing benign and malignant liver
tumors with precision, enhancing noninvasive diagnostic capabilities. DCE-MRI has shown potential in
providing detailed predictions of microvascular invasion (MVI) and vascular encapsulation tumor clusters
(VETC) in HCC, attaining the
area under the curve (AUC) of 0.85 in internal validation [9]. The integration of clinical and radiomics models
in DCE - MRI can improve diagnostic performance, offering a non-surgical substitute to traditional biopsy
methods. However, variability in imaging protocols and the need for standardization remain challenges that
could affect the clinical application of DCE-MRI, Building upon these foundations, this review aims to
objectively examine the breakthrough and clinical implementation of dynamic contrast-enhanced MRI in
oncological imaging. By weaving togeather contemporary publications on DCE-MRI’s diagnostic, prognostic
and therapeutic tracking mchanisms involving numerous tumor categories such as liver, spinal and breast
cancerous conditions.
Technical Foundation of DCE-MRI
A valuable means of evaluating tumor vascularity by quantifying how contrast agents move through tissue
compartments is offered by (DCE-MRI) Dynamic contrast-enhanced MRI. Perfusion rate at which contrast
material transitions from the blood flow dynamics vessels into the extracellular matrix is described through one
key parameter, Ktrans (volume transfer constant), both blood flow and the permeability of vessel walls is
reflected by this rate. To calculate Ktrans the TTofts model, which underpins many DCE-MRI analyses, treats
tissue as two interacting compartments and uses an arterial input function. Higher tumor aggressiveness and
indication of how the tumor might respond to therapies are often linked to elevated Ktrans [15,16].
Kep (rate constant), another crucial metric, exposes how quickly the contrast agent returns from the extracellular
matrix back into the bloodstream. Kep is typically computed as Ktrans divided by Ve (the volume of the
extracellular space). It provides insights into the efficacy of contrast clearance and highlights differences in
vascular permeability between tumor types [17,18]. As shown in Figure 1, …
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 92
Fig 1. The values included are ktrans (0.398 min-1), Kep (1.407 min-1), Ve (0.98), and lauc observed in the
ROI was labelled as type IV on the basis of tissue contrast kinetics.
Formula:
……………(1)
The parameter Ve (extracellular volume fraction) measures the proportion of tissue volume occupied by the
extracellular space. Tumors with higher Ve values have larger extracellular areas due to factors like edema or a
loose extracellular matrix, features often seen in more aggressive malignancies [18,19].
Formula:
…………(2)
Collectively, Ktrans, Kep, and Ve help assess key features of tumor biology, including angiogenic, perfusion,
and vascular leakiness. High Ktrans and Kep values generally point to active tumor growth and rich blood
supply, while lower levels may suggest necrosis or poor vascularization. These parameters are critical in
predicting tumor behavior and potential treatment responses. Its important to note that accurate measurements
of these parameters can be affected by several technical factors. Motion during imaging, especially due to
breathing, can distort time-intensity data. Additionally, inconsistencies in contrast agent dosing, timing of scans,
or the choice of input functions can lead to variability in results. Standardizing imaging protocols and addressing
motion artifacts remain essential for reliable DCE-MRI analysis. [2,15]. As demonstrated in Figure 2
Formula:
…………..(3)
Fig 2. DCE-MRI outcomes in a 48-year-old male show low Ktrans, Kep, Ve, and iAUC values within a
constrained ROI, with a type III tissue contrast concentration curve.
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 93
METHODOLOGY
This review was carried out using an arranged literature search framework following established narrative review
methodology. Validated investigations examining dynamic contrast-enhanced MRI, pharmacokinetic modeling,
and quantitative parameters such as Ktrans, Kep, Ve and Vp in lever tumor cancer were recognized by PubMed,
Scopus, Google Scholar and ScienceDirect. The search included articles published from 1998 to 2025 using
multiple keywords Inclusion criteria were, Studies involving DCE-MRI quantitative analysis, articles using
Tofts, Extended Tofts, or advanced Pharmacokinetic models, studies recognized PK parameters with tumor
grade, vascularity, or clinical efficacy. Conference abstracts, non-pharmacokinetic studies, and articles lacking
quantitative assessment were excluded. Data from each study were gathered relating to methodology, PK
parameters used, diagnostic efficiency, strengths, and limitations to integrate patterns, verify gaps, and showcase
future research pathways.
Fig 3: Representing preferred reporting items for systematic reviews and meta-analyses (PRISMA) for
methodology
Diagnostic Applications of DCE-MRI
Diagnostic work of DCE-MRI in Differentiating Benign vs. Malignant Liver Lesions
In a regular clinical approach, it is rare to come across obstacles when trying to separate benign from malignant
soft tissue tumors utilizing standard MRI alone, due to intersecting morphological features and signal levels [20].
Latest progress, including diffusion-weighted imaging (DWI), specifically with intravoxel incoherent motion
(IVIM) structuring, has enabled improved distinction by examining water diffusion traits [21]. Parameters like
the apparent diffusion coefficient (ADC), pure diffusion coefficient (D), Pseudodiffusion coefficient (D), and
perfusion fraction (f) help distinguish true diffusion from perfusion-linked effects [22]. These techniques have
shown capability in improving tumor profiling. Supportively, DCE-MRI relates pharmacokinetic
simulating(e.g., Tofts model) to extract parameters like Ktrans, Kep, and Ve, and semi-quantitative metrics such
as Iauc, providing additional operational understandings to aid in tumor separation with studies reporting
significantly higher Ktrans values in HCC, reflecting its aggressive angiogenesis and greater microvascular
permeability [13]. with reported area under the curve (AUC) values of 1.0 during validation and 0.8 in final test
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 94
phases, confirming their potential for clinical application [23]. Support vector machine (SVM)-based approaches
have yielded detection rate and precision around 84% and 81% respectively, in classifying liver lesions,
reinforcing DCE-MRI’s diagnostic value.
Table 1. Summary of key methodological parameters from selected studies evaluating dynamic contrast-
enhanced MRI in oncologic imaging, including MRI field strength, contrast agent protocols, computational
analysis tools, pharmacokinetic models, assessed dynamic variables, and pre-/post-treatment timing.
Study
year
Tesl
a
MRI
Radiologic tracer
Computationa
l analysis
numerical
analysis
Reviewed
Dynamic
study
variables,
Duration
assessment:
Dynamic
MRI variables
Chu et
al [24]
.2013.
1.5
Tesl
a
MRI
Gd-DTPA was
injected
Intravenously at
0.1mmol/kg with
an infusion rate of
2.5 ml
NordicICE
version 2.3
(NordicNeurol
ab)
Tofts-
based
dual-
compartm
ent kinetic
framewor
k
Vp, Ktrans,
AUC, PE
Pre-therapy2-
115 days;
post-
treatment 0-
187 days
Spratt
et al
[25,
26].
1.5
Tesl
a
MRI
Gadolinium-DTPA
was administered
intravenously at
0.1 mmol/kg using
volumetric flow
rate at 2& e ml/s
NordicICE
(Nordic
Neurolab)
Tofts 2-
comparme
nt
pharmaco
kinetic
model
Vp, Ktrans,
Fifty-seven
days after -
thearpy (IQR,
51-62 days).
Kumar
et al.
[27],
2017
1.5
Tesl
a
MRI
Gd- contrast agent
0.1 mmol/kg at a
controlled
injection speed of
2-3 ml/s
NordicICE
version 2.3
(NordicNeuro
Lab)
Expanded
dual-
compartm
ent
adopted
from the
original
Tofts
approach
Vp, Ktrans
Pre and post
are not
specified
Lis et
al.
[28],
2017
1.5
Tesl
a
MRI
The contrast agent
gadolinium 0.1
mmol/kg fluid
transfer unit 2.5
ml/s
NordicICE
(NordicNeuro
Lab) and
MATLAB
(Mathworks)
Toft’s
pharmaco
kinetic
model
analysis
Vp, Ktrans
1 hour prior to
and following
therapy, as
opposed to
later phases
(range,51-504
days)
Chen
at al.
[15],
2021
3T
MRI
Gadopentetate
dimeglumine 0.1
mmol/kg,
throughput of 2
ml/s
GE ADW4.6
workstation
and GenIQ
software
Extended
Toft’s 2-
compartm
ent
pharmaco
kinetic
model
Ktrans, Kep,
Ve
1 weak prior
to therapy; 1/3
months post-
therapy
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 95
Chung et al . Assessed the clinical performance of conventional MRI in distinguishing benign from malignant
soft tissue tumors by assessing parameters like lesion depth, size, and signal heterogeneity. Their approach
yielded a detection rate of 64%, a precision of 85%, and an overall diagnostic precision of 77%. When compared
to findings from more recent studies employing multiparametric MRI, it is evident that incorporating multiple
imaging parameters enhances diagnostic precision. For instance, studies report improvements in sensitivity from
71% to 81 % while maintaining constant specificity (69% to 83%) and precision [25].
Hu et al. Has proven the possibility of utilizing radiomics-based machine learning to separate benign and
malignant soft tissue neoplasms. In their study, least absolute shrinkage and selection operator (LASSO) logistic
regression frameworks built on apparent diffusion coefficient (ADC) traits accomplished superior diagnostic
correctness (AUC=0.955, sensitivity = 83%, specificity = 100%) compared to those according to fat-suppressed
T2WI attributes
Fig 4:
Multiparametric MRI including IVIM-DWI and DCE-MRI indicates a malignant soft tissue neoplasm,
later confirmed as myeloid sarcoma. Imaging metrics such as ADC, IVIM, and perfusion parameters support
the diagnosis.
(AUC = 0.75). Pattern attributes such as mean ADC, HISTO-skewness, and GLCM-associated metrics recorded
slight variations in tissue heterogeneity. Significantly, benign tumors had negatively skewed ADC histograms,
while malignant lesions were positively skewed due to elevated cellularity. Their application of whole-tumor
volumetric division and replicable feature derivation using LIFEx software supports the methodological
effectiveness of radiomics in this background. These outcomes offer convincing proof that ADC-based
radiomics systems, when integrated with machine learning algorithms like LASSO, can efficiently identify
tumor types, providing potential clinical decision assistance tools with high accuracy [29].
Quantitative assessment from intravoxel incoherent motion DWI (IVIM-DWI) and dynamic contrast-enhanced
MRI (DCE-MRI) both suggested a malignant soft tissue neoplasm. Histopathological confirmation identified
the lesion as myeloid sarcoma. Panel D summarizes the results from the multiparametric MRI analysis, including
metrics such as ADC, IVIM, and perfusion-based parameters supporting the final diagnosis.
Key diffusion metrics, for example, the (ADC) apparent diffusion coefficient and the diffusion coefficient (D),
have demonstrated statistically appropriate differences among benign and malignant tumors [12]. While D is
expected to perform ADC due to its reduced sensitivity to perfusion-related effects and improved reflection of
true tissue diffusivity, receiver operating characteristic (ROC) analysis showed comparable diagnostic utility
between the two (AUC values of 0.752 for ADC and 0.742 for D) [30]. These observations align with findings
from Lim et al who noted significant differenciation using D but not ADC. Discrepencies may be differences in
imaging technique, including Rijswijk's use of a 1.5 T scanner with limited b-values in early IVIM-DWI [13,6]
Additionally, recent hypotheses propose that benign tumors might exhibit more heterogeneous or pronounced
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 96
microcapillary perfusion than malignant ones, contributing to the lack of significant differences in D and
perfusion fraction (f) between tumor types. This notion is supported by multiparametric MRI in improving soft
tissue tumor evaluation while also acknowledging the limitations of perfusion-sensitive indices in certain tumor
contexts [13]. Subset evaluation in the same study uncovered that IVIM-DWI was restricted in separating benign
from malignant myxoid tumors because of intersecting ADC figures, resulting in extracellular matrix
fluctuation[31]. Yet, DCE-MRI factors such as Ktrans and IAUC persisted significantly different even in these
difficult scenarios, validating the benefit of perfusion imaging in tumors with high myxoid information. This
showcases the relevance of adjusting the imaging method based on tumor structure [7].
Prediction of Microvascular invasion (MVI)
DCE-MRI combined with radiomics has advanced presurgical estimation of MVI in HCC. Models incorporating
intratumorally and peritumoral features have achieved validation AUCs as high as 0.879, understanding their
potential for improving surgical planning [22]. Other studies using contrast agents such as Gd-EOB-DTPA have
further highlighted the role of MRI-derived radiomics in guiding clinical decisions related to MVI risk [32].
[33], Feng conducted a study and set a goal to produce and authenticate an integrated internal tumor region and
tumor margin radiomics framework to forecast MVI before surgery for initial HCC cases that applied
gadolinium-ethoxy benzyl-diethylenetriamine (Gd-EOB-DTPA) improved MRI. One hundred and ten HCC
cases took part in this investigation. They had a presurgical and curative hepatectomy, Gd-EOB-DTPA advanced
MRI analysis, and 40.0 % & 38.2% of the subjects were Microvascular invasion affirmative forecasting
framework. By its precise specificity, sensitivity, and AUC, the integrated peritumoral & intratumoral radiomics
model facilitated doctors in forming reliable treatment moves pre-surgery [31]. Another article by [34]. by
implementation of radiomics-based nomograms for 356 cases with solitary HCC < 5cm, Microvascular Invasion,
and reappearance-free survival (RFS) was estimated. During verification, it has been observed that 0.879 was
the cohort AUC of the MVI nomogram by logistic regression observations and 0.920 by random forest. The
median RFS of 30.5 months and >96.9 months, in corresponding order, were observed in patients with an MVI.
Return was autonomously estimated by age, histologic MVI, alkaline phosphatase, and alanine aminotransferase,
with an AUC of 0.654 in the RFS endorsement cohort [21]. DCE-MRI combined with radiomics has advanced
the pre-surgical evaluation of MVI in HCC. Models incorporating intratumorally and peritumoral features have
achieved validation AUCs as high as 0.879, understanding their potential for improving surgical planning [31].
Other studies using contrast agents such as Gd-EOB-DTPA have further highlighted the role of MRI-derived
radiomics in guiding clinical decisions related to MVI risk [21].
Another study conducted by Kumar et al. In which he investigated the differentiation in duration to reappearance
identification between present time MRI and DCE-MRI. Localized reappearance about 18 months prior to
traditional MRI in five patients was detected by DCE-MRI (mean ± standard deviation, 6.6 ± 6.8 months) [28].
revealed that the Vp variable could identify effective reaction to therapy within 10 days post-intervention, which
was likely too early than traditional MRI. DCE-MRI varies subsequent to RT could anticipate tumor reaction in
less than 6 months, to evaluate the point of tumor steadiness that is approximately half the period essential by
standard MRI.
DCE-MRI in other Cancers
Beyond liver cancers, DCE-MRI has been applied in breast cancer, where radiomics models based on DCE
features have achieved AUCs of 0.89 for differentiating benign and malignant lesions using SVM classifiers. In
cervical cancers, DCE-MRI's semi-quantitative parameters have shown strong sensitivity and specificity for
malignancy detection, although some limitations persist in distinguishing among all tumor types [35].
DCE-MRI in the evaluation of spinal Metastases
Spinal metastases affect 40-70% of advanced cancer patients and are a major source of morbidity due to pain,
paralysis, and vertebral fractures. Radiotherapy is a common treatment; however, conventional methods have
shown limited ability to assess post-treatment MRI changes, as nearly half of lesions may appear unchanged
despite clinical response. DCE-MRI has emerged as a superior modality in this context, offering kinetic
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 97
parameters such as Ktrans, Ve, and Vp, which reflect changes in vascular permeability and interstitial fluid
space. These biomarkers are highly sensitive to angiogenic and cytotoxic responses, allowing earlier detection,
like RECIST. Several studies report that DCE-MRI can distinguish responders from non-responders at much
earlier time points after stereotactic body radiotherapy (SBRT) or CyberKnife therapy. This has important
implications for guiding timely salvage therapies and optimizing patient outcomes. However, small sample sizes
remain a limiting factor in many spinal metastasis DCE studies, indicating the need for larger-scale prospective
trials. [14,21]
Therapeutic Monitoring with DCE-MRI
Fluctuation in DCE-MRI criteria and Analysis of tumor outcome after Radiation Therapy
Numerous studies have explored the potential of DCE-MRI parameters particularly, Ktrans, Vp, and Ve, to
monitor tumor response to radiation therapy (RT). In a study by Spratt et al.,2016 a marked reduction in ktrans
was observed approximately two months following stereotactic body radiotherapy (SBRT). Specifically, up to
75% of pathologies showed a decline in mean ktrans, with some tumors demonstrating a reduction of up to 92%.
Interestingly, in one patient who experienced local recurrence, ktrans values actually increased. However, due
to a limited sample size, statistical comparison between responders and non-responders were not feasible. Kumar
et al. provided more definitive results, reporting a significant difference in ktrans reduction among treatment
effective and ineffective groups (-66% vs. -7%, p + 0.01), with no local recurrence observed in patients who
achieved the more substantial ktrans drop [27].Lis et al. evaluated early responses in six patients with soial
metastasis undergoing high-dose image-guided radiotherapy (HD IGRT). They found a median reduction in
ktrans from 4.84 to 2.3 within one-hour post-treatment. No progression was observed during a follow-up period
extending over a period of 839 days [15] Similarly, documented that responders demonstrated an average ktrans
reduction of -32.6%, while non-responders demonstrated an average ktrans reduction of -32.6%, while non-
responders showed a median increase of +20.4% (p + 0.001) . In contrast, (Chu et al.,2013) did not observe
statistically significant changes in ktrans following RT (p = 0.48), possibly due to differences in timing or
methodology [31].
(VP) For the plasma volume parameter , Li et al, found a rapid 65.2% reduction in median Vp one-hour post-
HD IGRT, a finding mirrored by Spratt et al, who documented a 58.7% decrease in spinal sarcomas post-SBRT.
Chu et al. highlighted the predictive value of Vp, showing a notable changes in Vp changes among treatment
effective groups (-65.66%) and ineffective groups (+145.27%+206.79%), while Kumar et al. corroborated this
with a -76% versus +30% difference (p = 0.01) [30]. (VE) In contrast, this parameter showed inconsistent trends.
Vellayappan’s study reported no statistically significant changes in Ve values over time post-SBRT [36], found
a mean increase of 161.9% on Ve five weeks post-RT. Despite this, no marked disparity were observed among
effective treatment groups and non-effective groups in Ve levels or percentage changes.[37]. These findings that
Ktrans and Vp are more reliable indicators of early therapeutic response to RT in liver and metastatic tumors
than Ve. However, variations across studies related to timing, tumor type, and imaging protocols highlight the
need for standardized methodologies to fully leverage DCE-MRI for treatment monitoring.
DCE-MRI, Gd -DTPA, gadolinium-diethylenetrimine Penta acetic acid; Vp proportion of tissue volume
occupied by blood plasma; Ktrans, constant representing the movement rate among blood plasma and
extracellular matrix; AUC, total signal intensity change over time; PE-maximum contrast enhancement
observed; IQR, interquartile span; Ve, extravascular extracellular space volume fractions; (PS) permeability
surface product; CT, RT, radiation therapy[38].
Comparative Performance of DCE-MRI
DCE-MRI offers several advantages over conventional imaging modalities. It provides higher sensitivity than
ultrasound, which is known for its lower sensitivity and operator dependency [35]. Compared to CT, DCE-MRI
avoids ionizing radiation and offers superior specificity [31]. Moreover, it mitigates motion artifacts seen in
standard MRI and improves the detection of smaller lesions, which is crucial for early diagnosis [33]. DCE-MRI
is an important tool in assessing therapeutic response. Changes in Ktrans after treatments, particularly for HCC,
can reflect shifts in tumor blood flow and permeability, offering early indications of response to ablative or
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 98
systemic therapies [34]. Significant reductions in Ktrans have been correlated with superior progression-free
survival (PFS) and overall survival (OS), showcasing its prognostic value [39]. Quantitative DCE-MRI models
also contribute to predicting microvascular invasion in early-stage HCC, aiding in the recognition of patients at
greater vulnerability of recurrence [34]. Although data on prostate cancers are more limited, DCE-MRI is widely
recognized for its broader role in monitoring vascularity and treatment responses across various tumor types
[40].
Table 2. Output evaluation among FS-T2WI-based and ADC-based Rdiomics system of identifying Soft tissue
Neoplasms
In a straightforward analysis study, Hu et al. assessed radiomics-derived machine learning frameworks obtained
from fat-suppressed T2-weighted imaging (FS-T2WI) and apparent diffusion coefficient (ADC) maps to
differentiate benign and malignant soft tissue neoplasms [29]. Utilizing LASSO-logistic regression, the ADC–
grounded design exhibited meaningfully superior diagnostic capability, reaching an AUC of 0.955 in the
verification cohort, with a sensitivity of 83%, a specificity of 100% and overall accuracy of 91.7%. In contrast,
the FS-T2WI framework generated reduced sensitivity (55%) and accuracy (70.8%), even though preserving
high specificity. The variation in AUC among the models was quantitatively notable (P=0.017), highlighting the
more effective identifying power of ADC extracted pattern traits. These outcomes revealed the diagnostic benefit
of diffusion-based radiomics models over conventional T2-weighted imaging, specifically when volumetric
examination and higher-level texture indicators are integrated. An overview of these outcomes metrics is
displayed in Table 3 [29].
Advances in Radiomics, AI, and Technical Innovations
Radiomics techniques that extract detailed quantitative features from DCE-MRI have enhanced clinical decision-
making. For example, models predicting microvascular invasion in HCC using DCE-MRI radiomic features
have accomplished AUCs of 0.868 in the learning phase and 0.857 in confirmation sets, underscoring their
diagnostic promise [41]. Although substantial progress has been made in classifying and standardizing radiomic
biomarkers, there remains a noticeable lack of emphasis on the careful palining and execution of radiomic studies
focused on imaging diagnostic signal breakthrough, while analyzing outcomes, many released investigations
exhibit methodological errors or fall short in providing adequate methodological transparency, hindering the
reader's ability to contextualize findings [42,43]. Drawing on recent collaborative experiences in radiomics
research and peer review roles for journals such as Radiology, key design and statistical considerations have
been identified [16]. This decision does not aim to deliver a comprehensive overview of technical radiomic
Parameter
--
P=0.017 (vs.FS-T2WI)
Statistical Significance
Accuracy
0.750
0.955
Accuracy
70.8%
91.7%
Specificity
96.0%
100.0%
Sensitivity
55.0%
83.3%
AUC (Validation)
FS-T2WI Model
ADC-Based Model
Feature Types used
Texture features (GLCM,
GLZLM) from FS-T2WI
Texture+Histogram
(e.g.,HISTO-skewness) from
ADC
Modeling method
LASSO- Logistic Regression
LASSO-Logistic Regression
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 99
features such as grayscale level or bin width, which are extensively reviewed elsewhere, nor does it attempt to
catalog all possible sources of bias. Instead, the goal is to highlight frequent pitfalls in radiomic study design
and propose practical strategies to mitigate them. By doing so, the field can move toward generating robust,
reproducible, and clinically relevant outcomes capable of meaningfully advancing patient care [44].
Fig 5: Flowchart of radiomics and machine learning.
Radiomics demands a huge quantity of superior-quality information for assessment. Easily, such databases are
not accessible. Most of the existing data collections are retrospective and hence do not have sufficient data on
the applicability of outcomes. There is a requirement for prospective data collection. The association between
diverse radiomic traits can be acquired by segmentation. Furthermore, the rapid evolution of automated
techniques retrieving extensive-scale numerical attributes from medical imaging has led to a significant surge in
radiomics-related publications .
Furthermore, the rapid evolution of automated techniques retrieving extensive-scale numerical attributes from
medical imaging has led to a significant surge in radiomics-related publications [45]. These studies often aim to
leverage combinations of imaging features for tasks such as disease diagnosis, prognosis, therapy planning, or
other decision-support functions. Rdiomics refers not to the features themselves, ranging from conventional
metrics like Hounsfield units to more complex textural and machine-learned parameters [46]. Despite its
promise, a critical concern is that only a minor fraction of these measurable visual diagnostics achieve clinical
translation. To date, no high-throughput radiomic signature has gained widespread clinical adoption. This
underscores the importance of recognizing potential methodological barriers, including variability in study
designs. Identifying and overcoming these hidden challenges is essential to realizing the full clinical potential of
radiomics [47].
Although its capabilities, radiomics has not yet attained broad clinical incorporation. Moskowitz et al. Note that
no high-processing-rate radiomic patterns are presently in standard clinical use. This lack of implementation is
mostly credited to irregular confirmation, a lack of normalization, and the existence of unsettled technical and
statistical biases.
Moskowitz et al recognize that many radiomic research efforts are subject to structured biases that weaken
applicability. Frequent challenges involve bias when imaging elements are also used to determine outcomes,
verification bias due to targeted selection of validated cases, and spectrum bias when study cohorts do not
represent the wider clinical population. The picking of images to be employed, both for preparation and
confirming a radiomics framework, needs to be evaluated. Certain results, such as histologic diagnosis, are
merely analyzed for a group of cases, based in part on medical analysis of imaging outcomes. Constraining the
research to these patient image results in validiation bias (Table 1,) which is a data absence issue that might
produce production of sensitivity that are too inflated predictions, of specificity that are too low, or in rare
scenarios, lack of capacity in a straightforward manner predict sensitivity and specificity. A distinct study design
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 100
has been recommended to prevent this bias, and there are numerous approaches put forward to bias-adjustment
ways when validation bias is considered inevitable [48].
Excessive fitting continues to be a major challenge in radiomics simulation. It happens when models are fitted
too closely to specific datasets, detecting noise rather than the true signal, thus lowering general reliability. This
issue, combined with uncorrected multiple testing where hundreds of features are explored without proper
statistical safeguards, frequently inflates performance metrics, as highlighted by Moskowitz et al Feature values
in radiomics are highly sensitive to image acquisition parameters, scanner types, and segmentation approaches.
Moskowitz et al. emphasize that even identical features extracted from the same anatomical region can vary
significantly across platforms or acquisition protocols, undermining reproducibility and cross-institutional
comparability. To ensure clinical relevance and reproducibility, rigorous study design and transparent reporting
are essential. As recommended by Moskowitz et al., adherence to guidelines such as STARD for diagnostic
accuracy, TRIPOD for prediction modeling, and REMARK for tumor markers can improve the quality of
radiomic research and facilitate eventual clinical adoption.
Table 3. Repeated contributors of deviation and Bias in Radiomic studies
Type (Study design)
Description
Spectrum bias
(Ransohoff et al.,)
[49].
Study are not fully representative of the population of interest
Example; model developed using only extreme cases(eg, very sick and/or
very healthy idividuals)
Verfication bias
(Begg et al.,1988)
[50].
Analysis only includes cases where the outcomes is axcertained, which is
nonrepresentative subset of the population of interest
Example: Only including patients with biopsies where the decision to
biopysy to determined based on imaging.
Incorporation bias
(Zhou et al.,2002)
[51].
Final result uses data from the images beings examined, Example;
Predicting the outcomes from CT images where the outcome is defined by
radiologists from CT imaging.
Software variability
(Fornacon et al.,2020)
[52].
Feature measurement of the same region of interest in the same scan can
give different results Example; custom designed attributes, computed on a
distinct platform, or with a different version of the same software, can have
different values despite compliance with accepted standards.
Operator variability
(Pavic et al., 2018)
[53].
Manual or semiautomated segmentation affects feature measurement
Example Inter-and intraoperator variablility exists in manual contours; this
variability is also influenced by the disease site and existing clinical
contour guidelines .
Bias due to overfitting
(Hawkins et al.,2004)
[54].
Model captures spurious association in the training data, in addition to
assosciation that would be replicated in similar data sets. Example A model
captures random variation (Noise) in the training data and appears to
perform well but does not work well in independent validation data.
Bias from exclusion
of indeterminate or
missing feature data
(Begg et al.,) [50].
Ignoring images with missing features measurements in analyses might
result in to biased assessments of the charecteristics and the algorithims
performance, as well as decreased generalizability of the algorithim .
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 101
To overcome these challenges, recent efforts have focused on improving study design, quality control, and
transparency in radiomics research. A robust radiomics study must incorporate sound feature selection,
appropriate modeling methodology, and comprehensive validation. Feature selection should be data-driven,
incorporating dimensionality reduction techniques to eliminate non-robust or redundant variables. Modeling
often relies on a single machine learning technique, but the use of multiple algorithms can enhance
generalizability. Crucially, the validation process should include discrimination and calibration metrics,
supported by bootstrapping or cross-validation to reduce sampling bias [56]. To standardize methodological
rigor, the radiomics quality score (RQS) was developed, guiding researchers to report protocols, QA procedures,
and justification for model decisions. RQS also encourages full transparency in prediction model development,
complementing frameworks like TRIPOD, which mandate complete and unbiased reporting of multi-variable
prediction models. These frameworks are essential for reducing bias, improving replicability, and facilitating the
clinical adoption of radiomics-based tools [57].
DISCUSSION
(DCE-MRI) Dynamic contrast-enhanced MRI continues to gain traction as a powerful imaging modality for non
invasively assessing liver tumors, offering insights into perfusion dynamics, vascular permeability, and
extracellular volume. Its strength lies in the ability to acquire images continuously, often during free-breathing,
with motion corrected reconstruction that achieves high temporal resolution typically within 6-second intervals
[15]. Commercially available post-processing tools facilitate this process by incorporating motion correction
algorithms with adjustable translational and reproducibility [58]. Quantitative pharmacokinetic parameters
derived from DCE-MRI, including Ktrans, Kep, and Ve, serve as critical indicators of tumor behavior and
treatment response. These metrics are estimated on a voxel-wise framework utilizing constrained nonlinear
modeling methods, often supported by models that account for additional variables such as arterial and venous
plasma flow, extracellular volume, and delay times. Such modeling techniques, using Lavenberg-Marquardt
optimization, are reinforced with constraints to maintain physiologic plausibility [59].
In clinical settings, these parameters offer significant diagnostic values, with evidence showing that they can
detect therapeutic effects, recurrence, or treatment failures earlier than conventional imaging techniques. For
example, post-radiotherapy reductions in ktrans and Vp have consistently been observed among treatment
responders, suggesting a correlation between these perfusion metrics and therapeutic efficacy [60]. Despite these
benefits, several challenges hinder DCE-MRI's universal adoption. Tumor heterogeneity, inconsistent imaging
follow-ups, and technical complexity can affect parameter reliability [61]. Additionally, achieving diagnostically
sufficient image quality can be cumbersome, often requiring repeat contrast-enhanced scans, adding burden to
both patients and healthcare workflows. Nevertheless, DCE-MRI remains an essential tool, especially when
paired with systematic statistical evaluation and interobserver agreement. Its potential for early response
prediction and precision monitoring reaffirms its emerging role in oncology imaging, although alignment and
refinement of imaging parameters remain areas for ongoing development [47].
The advanced clinical utility of Dynamic Contrast-enhanced-MRI in liver cancer, upcoming research should
target many key areas. Standardization of parameters and acquisition protocols is essential to reassure routine
use of DCE-MRI metrics, particularly for distinguishing between liver pathologies like (HCC) hepatocellular
carcinoma and (HCA) hepatocellular adenoma. Longitudinal studies tracking DCE-MRI parameter changes over
time could offer precious insights into tumor progression and treatment response, supporting the development
of predictive models for patients' outcomes. Integrating functional MRI techniques alongside DCE-MRI could
enhance spatial mapping of liver function and inform more personalized treatment plans [13]. Technical
innovations, including advanced K-space undersampling, golden-angle sampling, and training radiologists
without excessive use of contrast agents [12]. There is also growing interest in deep learning frameworks to
Optimistic
Performance bias
(Harrell et al.,1996)
[55].
Evaluating the algorithims on the same data that was used to build or
optimize the algorithim.
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 102
improve parameter estimation accuracy and reduce computational demands, thereby enhancing diagnostic
confidence [13]. Developing cost-effective solutions, such as streamlined protocols and affordable contrast
agents, alongside telemedicine integration protocols, DCE-MRI accessibility, especially in low-resource settings
[62]. These research directions aim to enhance both the diagnostic power and clinical practicality of DCE-MRI
in liver cancer care.
CONCLUSION
Recent reviews consistently highlight the substantial diagnostic significance of Dynamic Contrast Enhanced
Magnetic Resonance Imaging in the diagnosis, observation, and characterization of liver tumors. DCE-MRI has
proven more effective than standard methods of imaging modalities in categorizing benign from malignant liver
lesions, evaluating microvascular invasion, and assessing treatment response. The quantitative parameters it
provides, such as Ktrans, Kep, and Ve, offer critical insights into tumor vascularity and permeability that
supports more informed clinical decision-making. Although challenges like protocols variability, motion,
artifacts, and limited standardization remain, the integration of radiomics and artificial intelligence with DCE-
MRI is seen as a promising direction for improving diagnostic accuracy and patient outcomes. As ongoing
research continues to address these limitations, DCE-MRI is likely to make a highly demanding, vital
contribution to personalized liver cancer care.
Ethical Statement
None of the authors have conducted any research on humans or animals for this paper.
Conflicts of Interest
The authors declare no conflicts of interest related to this work.
Data Availability Statement
Since this study did not create or analyse any new data, data sharing is not applicable to this article.
FUNDING
The author(s) declare that no financial support was received for the research, authorship, and/or publication of
this article.
Author Contribution Statement
Faizan Farooq: original draft, Review and editing , Mohit Sharma: Supervision, Visualization. Khursheed
Ahmed: Conceptualization, Formal Analysis, Srishti Bhardwaj: Formal Analysis and Proofreading.
REFERENCES
1. Shao YY, et al. Lenalidomide as second‐line therapy for advanced hepatocellular carcinoma: exploration
of biomarkers for treatment efficacy. Alimentary Pharmacology & Therapeutics. 2017 Oct;46(8):722-
30.
2. Chen BB, Shih TT. DCE-MRI in hepatocellular carcinoma-clinical and therapeutic image biomarker.
World journal of gastroenterology: WJG. 2014 Mar 28;20(12):3125.
3. Zhang J, et al. Assessment of tumor treatment response using active contrast encoding (ACE)-MRI:
Comparison with conventional DCE-MRI. PLoS One. 2020 Jun 10;15(6):e0234520.
4. O'Connor JP, Jackson A, et al. DCE-MRI biomarkers in the clinical evaluation of antiangiogenic and
vascular disrupting agents. British journal of cancer. 2007 Jan;96(2):189-95.
5. Song J, Gu Y, Du T, Liu Q. Analysis of quantitative and semi-quantitative parameters of DCE-MRI in
differential diagnosis of benign and malignant cervical tumors. American Journal of Translational
Research. 2021 Nov 15;13(11):12228.
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 103
6. Notley SV, Thacker NA, et al. Reliability of Tumour Classification from Multi-Dimensional DCE-MRI
Variables using Data Transformations. arXiv preprint arXiv:2303.10014. 2023 Mar 17.
7. Lee SK, Jee WH, Jung CK, Chung YG. Multiparametric quantitative analysis of tumor perfusion and
diffusion with 3T MRI: differentiation between benign and malignant soft tissue tumors. The British
journal of radiology. 2020 Nov 1;93(1115):20191035.
8. Mourad MA, Higazi MM. MRI prognostic factors of tongue cancer: potential predictors of cervical
lymph nodes metastases. Radiology and Oncology. 2019 Mar 3;53(1):49
9. Gao A, et al.,Applying dynamic contrast-enhanced MRI tracer kinetic models to differentiate benign and
malignant soft tissue tumors. Cancer Imaging. 2024 May 21;24(1):64.
10. Shao YY, et al. Lenalidomide as second‐line therapy for advanced hepatocellular carcinoma: exploration
of biomarkers for treatment efficacy. Alimentary Pharmacology & Therapeutics. 2017 Oct;46(8):722-
30.
11. Yang J, Dong X, et al. Radiomics model of dynamic contrast-enhanced MRI for evaluating vessels
encapsulating tumor clusters and microvascular invasion in hepatocellular carcinoma. Academic
Radiology. 2025 Jan 1;32(1):146-56.
12. Wassenaar NP, et al. Optimizing pseudo‐spiral sampling for abdominal DCE MRI using a digital
anthropomorphic phantom. Magnetic Resonance in Medicine. 2024 Nov;92(5):2051-64.
13. Alareer HS, Taher HJ, Abdullah AD. Dynamic contrast enhanced-MRI and apparent diffusion coefficient
quantitation for differentiate hepatocellular carcinoma from hepatocellular adenoma. Asian Pacific
Journal of Cancer Prevention: APJCP. 2024;25(3):931..
14. Wang H, Guo S. Solitary hypovascular hepatic nodules: comparison and differentiation between
peripheral nodular cholangiocarcinoma and atypical liver hemangioma with MRI. Translational Cancer
Research. 2023 Aug 25;12(9):2308.
15. Chen Y et al., Use of dynamic contrast-enhanced MRI for the early assessment of outcome of CyberKnife
stereotactic radiosurgery for patients with spinal metastases. Clinical Radiology. 2021 Nov 1;76(11):864-
e1.
16. Granata V, Grassi R, et al., La Porta M, Brunese MC, De Muzio F, Ottaiano A. Diagnostic evaluation
and ablation treatments assessment in hepatocellular carcinoma. Infectious Agents and Cancer. 2021 Jul
19;16(1):53.
17. Minutoli F et al,. Effect of granulocyte colony-stimulating factor on bone marrow: evaluation by
intravoxel incoherent motion and dynamic contrast-enhanced magnetic resonance imaging. La radiologia
medica. 2020 Mar;125(3):280-7.
18. Chen BB, Shih TT. DCE-MRI in hepatocellular carcinoma-clinical and therapeutic image biomarker.
World journal of gastroenterology: WJG. 2014 Mar 28;20(12):3125.
19. Spratt DE, et al. Early magnetic resonance imaging biomarkers to predict local control after high dose
stereotactic body radiotherapy for patients with sarcoma spine metastases. The Spine Journal. 2016 Mar
1;16(3):291-8.
20. Kransdorf MJ, Jelinek JS , et al. Soft- Tissue masses: diagnosis using MR imaging. AJR Am J
Roentgenol 1989; 153: 541–
21. van Rijswijk CS, Kunz P, et al. Diffusion‐weighted MRI in the characterization of soft‐tissue tumors.
Journal of Magnetic Resonance Imaging: An Official Journal of The International Society For Magnetic
Resonance In Medicine. 2002 Mar;15(3):302-7.
22. Wu H, Zhang S, Liang C, Liu H, Liu Y, Mei Y, et al. Intravoxel incoherent motion MRI for the
differentiation of benign, intermediate, and malignant solid soft- tissue tumors. J Magn Reson Imaging
2017; 46: 1611–8.
23. Liu X, et al.,Multi-modality radiomics nomogram based on DCE-MRI and ultrasound images for benign
and malignant breast lesion classification. Frontiers in Oncology. 2022 Dec 2;12:992509.
24. Chung et al., MRI to differentiate benign from malignant soft-tissue tumours of the extremities: a
simplified systematic imaging approach using depth, size and heterogeneity of signal intensity. The
British journal of radiology. 2012 Oct 1;85(1018):e831-6.
25. Chu S, et al., Holodny AI. Measurement of blood perfusion in spinal metastases with dynamic contrast-
enhanced magnetic resonance imaging: evaluation of tumor response to radiation therapy. Spine. 2013
Oct 15;38(22):E1418-24.
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 104
26. Spratt DE, et al. Early magnetic resonance imaging biomarkers to predict local control after high dose
stereotactic body radiotherapy for patients with sarcoma spine metastases. The Spine Journal. 2016 Mar
1;16(3):291-8.
27. Kumar KA et al. A pilot study evaluating the use of dynamic contrast-enhanced perfusion MRI to predict
local recurrence after radiosurgery on spinal metastases. Technology in Cancer Research & Treatment.
2017 Dec;16(6):857-65.
28. Lis E, et al,.Dynamic contrast-enhanced magnetic resonance imaging of osseous spine metastasis before
and 1 hour after high-dose image-guided radiation therapy. Neurosurgical focus. 2017 Jan 1;42(1):E9.
29. Hu P, Chen L, Zhou Z. Machine learning in the differentiation of soft tissue neoplasms: comparison of
fat-suppressed T2WI and apparent diffusion coefficient (ADC) features-based models. Journal of Digital
Imaging. 2021 Oct;34(5):1146-55.
30. Vellayappan AP, et al. Quantifying the changes in the tumour vascular micro-environment in spinal
metastases treated with stereotactic body radiotherapy-a single arm prospective study. Radiology and
Oncology. 2022 Dec 13;56(4):525.
31. Lim HK, Jee WH, et al, incoherent motion diffusion-weighted MR imaging for differentiation of benign
and malignant musculoskeletal tumours at 3 T. The British journal of radiology. 2018 Feb
1;91(1082):20170636
32. Sigmund EE,et al. Intravoxel incoherent motion imaging of tumor microenvironment in locally advanced
breast cancer. Magnetic resonance in medicine. 2011 May;65(5):1437-47
33. Feng ST, et al.,prediction of microvascular invasion in hepatocellular cancer: a radiomics model using
Gd-EOB-DTPA-enhanced MRI. European radiology. 2019 Sep 1;29(9):4648-59.
34. Chong HH et al.,Multi-scale and multi-parametric radiomics of gadoxetate disodium–enhanced MRI
predicts microvascular invasion and outcome in patients with solitary hepatocellular carcinoma≤ 5 cm.
European Radiology. 2021 Jul;31(7):4824-38.
35. Qiao M, Liu C, et al. Breast tumor classification based on MRI-US images by disentangling modality
features. IEEE Journal of Biomedical and Health Informatics. 2022 Jan 4;26(7):3059-67.
36. Lee JH, Yoo GS et al,. Diffusion-weighted and dynamic contrast-enhanced magnetic resonance imaging
after radiation therapy for bone metastases in patients with hepatocellular carcinoma. Scientific Reports.
2021 May 17;11(1):10459.
37. Fotouhi M, et al.,Assessment of LI-RADS efficacy in classification of hepatocellular carcinoma and
benign liver nodules using DCE-MRI features and machine learning. European Journal of Radiology
Open. 2023 Dec 1;11:100535.
38. Qiao M, Li C, Suo S, et al. Breast DCE-MRI radiomics: a robust computer-aided system based on
reproducible BI-RADS features across the influence of datasets bias and segmentation methods.
International journal of computer assisted radiology and surgery. 2020 Jun;15(6):921-30
39. Guo R, Lu G, Qin B, Fei B. Ultrasound imaging technologies for breast cancer detection and
management: a review. Ultrasound in medicine & biology. 2018 Jan 1;44(1):37-70.
40. Pinsky RW, Helvie MA. Mammographic breast density: effect on imaging and breast cancer risk. Journal
of the National Comprehensive Cancer Network. 2010 Oct 1;8(10):1157-65.
41. Jacobs MA, et al.,. Benign and malignant breast lesions: diagnosis with multiparametric MR imaging.
Radiology. 2003 Oct;229(1):225-32.
42. Segal E, Sirlin CB, et al. Decoding global gene expression programs in liver cancer by noninvasive
imaging. Nature biotechnology. 2007 Jun;25(6):675-80.
43. Zhu Y, et al. Bi-regional dynamic contrast-enhanced MRI for prediction of microvascular invasion in
solitary BCLC stage A hepatocellular carcinoma. Insights into Imaging. 2024 Jun 18;15(1):149.
44. Ronot M, Nahon P et al. Screening of liver cancer with abbreviated MRI. Hepatology. 2023 Aug
1;78(2):670-86..
45. Zwanenburg A, et al. The image biomarker standardization initiative: standardized quantitative radiomics
for high-throughput image-based phenotyping. Radiology. 2020 May;295(2):328-38.
46. Kessler LG, et al., QIBA Terminology Working Group. The emerging science of quantitative imaging
biomarkers terminology and definitions for scientific studies and regulatory submissions. Statistical
methods in medical research. 2015 Feb;24(1):9-26.
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025
www.ijltemas.in Page 105
47. Park JE, Kim D et al. Quality of science and reporting of radiomics in oncologic studies: room for
improvement according to radiomics quality score and TRIPOD statement. European radiology. 2020
Jan;30(1):523-36.
48. Moskowitz CS, et al. Radiomic analysis: study design, statistical analysis, and other bias mitigation
strategies. Radiology. 2022 Aug;304(2):265-73.
49. Ransohoff DF, Feinstein AR. Problems of spectrum and bias in evaluating the efficacy of diagnostic
tests. N Engl J Med 1978;299(17):926–930.
50. Begg CB, McNeil BJ. Assessment of radiologic tests: control of bias and other design considerations.
Radiology 1988;167(2):565–569.
51. Zhou XH, et al. In: Statistical Methods in Di agnostic Medicine. New York, NY: Wiley-Interscience,
2002; 73–74.
52. Fornacon-Wood I, Mistry H, Ackermann CJ, et al. Reliability and prog nostic value of radiomic features
are highly dependent on choice of fea ture extraction platform. Eur Radiol 2020;30(11):6241–6250.
53. Pavic M, Bogowicz M, Würms X, et al. Influence of inter-observer de lineation variability on radiomics
stability in different tumor sites. Acta Oncol 2018;57(8):1070–1074.
54. Hawkins DM, et al. (2004). The problem of overfitting. Journal of Chemical Information and Computer
Sciences, 44(1), 1–12.
55. Harrell FE Jr, et al. (1996). Multivariable prognostic models: issues in developing models, evaluating
assumptions and adequacy, and measuring and reducing errors. Statistics in Medicine, 15(4), 361–387
56. Collins GS, et al. (2015). Transparent reporting of a multivariable prediction model for individual
prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ, 350, g7594.
57. Gillies RJ, et al. (2016). Radiomics: images are more than pictures, they are data. Radiology, 278(2),
563–577.
58. Tadimalla S, Wang W, et al. Role of functional MRI in liver SBRT: Current use and future directions.
Cancers. 2022 Nov 28;14(23):5860.
59. Mulyadi R, et al. (2023). Dynamic contrast-enhanced magnetic resonance imaging parameter changes as
an early biomarker of tumor responses following radiation therapy in patients with spinal metastases: a
systematic review. Radiation Oncology Journal, 41(4), 225.
60. Yoon JH, et al. (2022). Simultaneous evaluation of perfusion and morphology using GRASP MRI in
hepatic fibrosis. European Radiology, 32(1), 34–45.
61. Dejene EM, et al. (2023). Unified Bayesian network for uncertainty quantification of physiological
parameters in dynamic contrast enhanced (DCE) MRI of the liver. Physics in Medicine & Biology,
68(21), 215018.
62. Morin O, et al. (2018). A deep look into the future of quantitative imaging in oncology: a statement of
working principles and proposal for change. International Journal of Radiation Oncology Biology*
Physics*, 102(4), 1074–1082.
