Prospective clinical research of radiomics and deep learning in oncology: A translational review

The Application Status of Radiomics-Based Machine Learning in Intrahepatic Cholangiocarcinoma: Systematic Review and Meta-Analysis

BACKGROUND

Over the past few years, radiomics for the detection of intrahepatic cholangiocarcinoma (ICC) has been extensively studied. However, systematic evidence is lacking in the use of radiomics in this domain, which hinders its further development.

OBJECTIVE

To address this gap, our study delved into the status quo and application value of radiomics in ICC and aimed to offer evidence-based support to promote its systematic application in this field.

METHODS

PubMed, Web of Science, Cochrane Library, and Embase were comprehensively retrieved to determine relevant original studies. The study quality was appraised through the Radiomics Quality Score. In addition, subgroup analyses were undertaken according to datasets (training and validation sets), imaging sources, and model types.

RESULTS

Fifty-eight studies encompassing 12,903 patients were eligible, with an average Radiomics Quality Score of 9.21. Radiomics-based machine learning (ML) was mainly used to diagnose ICC (n=30), microvascular invasion (n=8), gene mutations (n=5), perineural invasion (PNI; n=2), lymph node (LN) positivity (n=2), and tertiary lymphoid structures (TLSs; n=2), and predict overall survival (n=6) and recurrence (n=9). The C-index, sensitivity (SEN), and specificity (SPC) of the ML model developed using clinical features (CFs) for ICC detection were 0.762 (95% CI 0.728-0.796), 0.72 (95% CI 0.66-0.77), and 0.72 (95% CI 0.66-0.78), respectively, in the validation dataset. In contrast, the C-index, SEN, and SPC of the radiomics-based ML model for detecting ICC were 0.853 (95% CI 0.824-0.882), 0.80 (95% CI 0.73-0.85), and 0.88 (95% CI 0.83-0.92), respectively. The C-index, SEN, and SPC of ML constructed using both radiomics and CFs for diagnosing ICC were 0.912 (95% CI 0.889-0.935), 0.77 (95% CI 0.72-0.81), and 0.90 (95% CI 0.86-0.92). The deep learning-based model that integrated both radiomics and CFs yielded a notably higher C-index of 0.924 (0.863-0.984) in the task of detecting ICC. Additional analyses showed that radiomics demonstrated promising accuracy in predicting overall survival and recurrence, as well as in diagnosing microvascular invasion, gene mutations, PNI, LN positivity, and TLSs.

CONCLUSIONS

Radiomics-based ML demonstrates excellent accuracy in the clinical diagnosis of ICC. However, studies involving specific tasks, such as diagnosing PNI and TLSs, are still scarce. The limited research on deep learning has hindered both further analysis and the development of subgroup analyses across various models. Furthermore, challenges such as data heterogeneity and interpretability caused by segmentation and imaging parameter variations require further optimization and refinement. Future research should delve into the application of radiomics to enhance its clinical use. Its integration into clinical practice holds great promise for improving decision-making, boosting diagnostic and treatment accuracy, minimizing unnecessary tests, and optimizing health care resource usage.

Evaluation of EGFR-TKIs and ICIs treatment stratification in non-small cell lung cancer using an encrypted multidimensional radiomics approach

BACKGROUND

Radiomics holds great potential for the noninvasive evaluation of EGFR-TKIs and ICIs responses, but data privacy and model robustness challenges limit its current efficacy and safety. This study aims to develop and validate an encrypted multidimensional radiomics approach to enhance the stratification and analysis of therapeutic responses.

MATERIALS AND METHODS

This multicenter study incorporated various data types from 506 NSCLC patients, which underwent preprocessing through anonymization methods and were securely encrypted using the AES-CBC algorithm. We developed one clinical model and three radiomics models based on clinical factors and radiomics scores (RadScore) of three distinct regions to evaluate treatment response. Additionally, an integrated radiomics-clinical model was created by combining clinical factors with RadScore. The study also explored the association between different EGFR mutations and PD-1/PD-L1 expression in radiomics biomarkers.

FINDINGS

The radiomics-clinical model demonstrated high performance, with AUC values as follows: EGFR (0.884), 19Del (0.894), L858R (0.881), T790M (0.900), and PD-1/PD-L1 expression (0.893) in the test set. This model outperformed both clinical and single radiomics models. Decision curve analysis further supported its superior clinical utility. Additionally, our findings suggest that the efficacy of EGFR-TKIs and ICIs therapy may not depend on detecting a singular tumor feature or cell type.

CONCLUSION

The proposed method effectively balances the level of evidence with privacy protection, enhancing the study's validity and security. Therefore, radiomics biomarkers are expected to complement molecular biology analyses and guide therapeutic strategies for EGFR-TKIs, ICIs, and their combinations.

Exploiting biochemical data to improve osteosarcoma diagnosis with deep learning

Early and accurate diagnosis of osteosarcomas (OS) is of great clinical significance, and machine learning (ML) based methods are increasingly adopted. However, current ML-based methods for osteosarcoma diagnosis consider only X-ray images, usually fail to generalize to new cases, and lack explainability. In this paper, we seek to explore the capability of deep learning models in diagnosing primary OS, with higher accuracy, explainability, and generality. Concretely, we analyze the added value of integrating the biochemical data, i.e., alkaline phosphatase (ALP) and lactate dehydrogenase (LDH), and design a model that incorporates the numerical features of ALP and LDH and the visual features of X-ray imaging through a late fusion approach in the feature space. We evaluate this model on real-world clinic data with 848 patients aged from 4 to 81. The experimental results reveal the effectiveness of incorporating ALP and LDH simultaneously in a late fusion approach, with the accuracy of the considered 2608 cases increased to 97.17%, compared to 94.35% in the baseline. Grad-CAM visualizations consistent with orthopedic specialists further justified the model's explainability.

Development and interpretation of a multimodal predictive model for prognosis of gastrointestinal stromal tumor

Gastrointestinal stromal tumor (GIST) is the most common mesenchymal original tumor in gastrointestinal (GI) tract and is considered to have varying malignant potential. With the advancement of computer science, radiomics technology and deep learning had been applied in medical researches. It's vital to construct a more accurate and reliable multimodal predictive model for recurrence-free survival (RFS) aiding for clinical decision-making. A total of 254 patients underwent surgery and pathologically diagnosed with GIST in The First Hospital of China Medical University from 2019 to 2022 were included in the study. Preoperative contrast enhanced computerized tomography (CE-CT) and hematoxylin/eosin (H&E) stained whole slide images (WSI) were acquired for analysis. In the present study, we constructed a sum of 11 models while the multimodal model (average C-index of 0.917 on validation set in 10-fold cross validation) performed the best on external validation cohort with an average C-index of 0.864. The multimodal model also reached statistical significance when validated in the external validation cohort (n = 42) with a p-value of 0.0088 which pertained to the recurrence-free survival (RFS) comparison between the high and low groups using the optimal threshold on the predictive score. We also explored the biological significance of radiomics and pathomics features by visualization and quantitative analysis. In the present study, we constructed a multimodal model predicting RFS of GIST which was prior over unimodal models. We also proposed hypothesis on the correlation between morphology of tumor cell and prognosis.

Non-invasive decision support for clinical treatment of non-small cell lung cancer using a multiscale radiomics approach

BACKGROUND

Selecting therapeutic strategies for cancer patients is typically based on key target-molecule biomarkers that play an important role in cancer onset, progression, and prognosis. Thus, there is a pressing need for novel biomarkers that can be utilized longitudinally to guide treatment selection.

METHODS

Using data from 508 non-small cell lung cancer (NSCLC) patients across three institutions, we developed and validated a comprehensive predictive biomarker that distinguishes six genotypes and infiltrative immune phenotypes. These features were analyzed to establish the association between radiological phenotypes and tumor genotypes/immune phenotypes and to create a radiological interpretation of molecular features. In addition, we assessed the sensitivity of the models by evaluating their performance at five different voxel intervals, resulting in improved generalizability of the proposed approach.

FINDINGS

The radiomics model we developed, which integrates clinical factors and multi-regional features, outperformed the conventional model that only uses clinical and intratumoral features. Our combined model showed significant performance for EGFR, KRAS, ALK, TP53, PIK3CA, and ROS1 mutation status with AUCs of 0.866, 0.874, 0.902, 0.850, 0.860, and 0.900, respectively. Additionally, the predictive performance for PD-1/PD-L1 was 0.852. Although the performance of all models decreased to different degrees at five different voxel space resolutions, the performance advantage of the combined model did not change.

CONCLUSIONS

We validated multiscale radiomic signatures across tumor genotypes and immunophenotypes in a multi-institutional cohort. This imaging-based biomarker offers a non-invasive approach to select patients with NSCLC who are sensitive to targeted therapies or immunotherapy, which is promising for developing personalized treatment strategies during therapy.

Exploring non-invasive precision treatment in non-small cell lung cancer patients through deep learning radiomics across imaging features and molecular phenotypes

BACKGROUND

Accurate prediction of tumor molecular alterations is vital for optimizing cancer treatment. Traditional tissue-based approaches encounter limitations due to invasiveness, heterogeneity, and molecular dynamic changes. We aim to develop and validate a deep learning radiomics framework to obtain imaging features that reflect various molecular changes, aiding first-line treatment decisions for cancer patients.

METHODS

We conducted a retrospective study involving 508 NSCLC patients from three institutions, incorporating CT images and clinicopathologic data. Two radiomic scores and a deep network feature were constructed on three data sources in the 3D tumor region. Using these features, we developed and validated the 'Deep-RadScore,' a deep learning radiomics model to predict prognostic factors, gene mutations, and immune molecule expression levels.

FINDINGS

The Deep-RadScore exhibits strong discrimination for tumor molecular features. In the independent test cohort, it achieved impressive AUCs: 0.889 for lymphovascular invasion, 0.903 for pleural invasion, 0.894 for T staging; 0.884 for EGFR and ALK, 0.896 for KRAS and PIK3CA, 0.889 for TP53, 0.895 for ROS1; and 0.893 for PD-1/PD-L1. Fusing features yielded optimal predictive power, surpassing any single imaging feature. Correlation and interpretability analyses confirmed the effectiveness of customized deep network features in capturing additional imaging phenotypes beyond known radiomic features.

INTERPRETATION

This proof-of-concept framework demonstrates that new biomarkers across imaging features and molecular phenotypes can be provided by fusing radiomic features and deep network features from multiple data sources. This holds the potential to offer valuable insights for radiological phenotyping in characterizing diverse tumor molecular alterations, thereby advancing the pursuit of non-invasive personalized treatment for NSCLC patients.

A multimodal radiomic machine learning approach to predict the LCK expression and clinical prognosis in high-grade serous ovarian cancer

We developed and validated a multimodal radiomic machine learning approach to noninvasively predict the expression of lymphocyte cell-specific protein-tyrosine kinase (LCK) expression and clinical prognosis of patients with high-grade serous ovarian cancer (HGSOC). We analyzed gene enrichment using 343 HGSOC cases extracted from The Cancer Genome Atlas. The corresponding biomedical computed tomography images accessed from The Cancer Imaging Archive were used to construct the radiomic signature (Radscore). A radiomic nomogram was built by combining the Radscore and clinical and genetic information based on multimodal analysis. We compared the model performances and clinical practicability via area under the curve (AUC), Kaplan-Meier survival, and decision curve analyses. LCK mRNA expression was associated with the prognosis of HGSOC patients, serving as a significant prognostic marker of the immune response and immune cells infiltration. Six radiomic characteristics were chosen to predict the expression of LCK and overall survival (OS) in HGSOC patients. The logistic regression (LR) radiomic model exhibited slightly better predictive abilities than the support vector machine model, as assessed by comparing combined results. The performance of the LR radiomic model for predicting the level of LCK expression with five-fold cross-validation achieved AUCs of 0.879 and 0.834, respectively, in the training and validation sets. Decision curve analysis at 60 months demonstrated the high clinical utility of our model within thresholds of 0.25 and 0.7. The radiomic nomograms were robust and displayed effective calibration. Abnormally high expression of LCK in HGSOC patients is significantly correlated with the tumor immune microenvironment and can be used as an essential indicator for predicting the prognosis of HGSOC. The multimodal radiomic machine learning approach can capture the heterogeneity of HGSOC, noninvasively predict the expression of LCK, and replace LCK for predictive analysis, providing a new idea for predicting the clinical prognosis of HGSOC and formulating a personalized treatment plan.

Radiomics under 2D regions, 3D regions, and peritumoral regions reveal tumor heterogeneity in non-small cell lung cancer: a multicenter study

PURPOSE

Lung cancer has significant genetic and phenotypic heterogeneity, leading to poor prognosis. Radiomic features have emerged as promising predictors of the tumor phenotype. However, the role of underlying information surrounding the cancer remains unclear.

MATERIALS AND METHODS

We conducted a retrospective study of 508 patients with NSCLC from three institutions. Radiomics models were built using features from six tumor regions and seven classifiers to predict three prognostically significant tumor phenotypes. The models were evaluated and interpreted by the mean area under the receiver operating characteristic curve (AUC) under nested cross-validation and Shapley values. The best-performing predictive models corresponding to six tumor regions and three tumor phenotypes were identified for further comparative analysis. In addition, we designed five experiments with different voxel spacing to assess the sensitivity of the experimental results to the spatial resolution of the voxels.

RESULTS

Our results demonstrated that models based on 2D, 3D, and peritumoral region features yielded mean AUCs and 95% confidence intervals of 0.759 and [0.747-0.771] for lymphovascular invasion, 0.889 and [0.882-0.896] for pleural invasion, and 0.839 and [0.829-0.849] for T-staging in the testing cohort, which was significantly higher than all other models. Similar results were obtained for the model combining the three regional features at five voxel spacings.

CONCLUSION

Our study revealed the predictive role of the developed methods with multi-regional features for the preoperative assessment of prognostic factors in NSCLC. The analysis of different voxel spacing and model interpretability strengthens the experimental findings and contributes to understanding the biological significance of the radiological phenotype.