Radiomics in Urolithiasis: Systematic Review of Current Applications, Limitations, and Future Directions

Ee Jean Lim, Daniele Castellani, Wei Zheng So, Khi Yung Fong, Jing Qiu Li, Ho Yee Tiong, Nariman Gadzhiev, Chin Tiong Heng, Jeremy Yuen‐Chun Teoh, Nithesh Naik, Khurshid R. Ghani, Kemal Sarıca, Jean de la Rosette, Bhaskar Somani, Vineet GauharJournal of Clinical Medicine 2022 · DOI

TL;DR

This systematic review explores the emerging role of radiomics in urolithiasis (kidney stones), demonstrating its potential to improve diagnosis, predict stone characteristics, and guide treatment outcomes. While still in its early stages, radiomics shows promise in differentiating stones from mimics, classifying stone types, and predicting treatment success, potentially leading to more personalized patient care.

What Problem Does This Paper Address?

Urolithiasis is a prevalent and often recurrent condition that poses diagnostic and management challenges. Differentiating kidney stones from similar-looking calcifications (like phleboliths) can be difficult, leading to unnecessary interventions. Furthermore, predicting stone composition, the likelihood of spontaneous passage, or the success of surgical procedures like PCNL and RIRS is crucial for effective treatment planning. Traditional methods often rely on subjective interpretation or invasive procedures. This paper addresses the need for more objective, non-invasive tools to enhance the diagnostic accuracy and therapeutic decision-making in urolithiasis management.

Methods

This paper is a systematic review of existing literature on radiomics applications in urolithiasis. The authors conducted an electronic literature search across PubMed, EMBASE, and Scopus up to March 2022, adhering to PRISMA guidelines for scoping reviews. They included studies that utilized radiomics for diagnostics, treatment prediction, or therapeutics in any form of urolithiasis. The review focuses on analyzing the findings of these studies, categorizing their applications, and evaluating their reported outcomes.

Datasets

The paper reviews studies that utilized various datasets, primarily derived from computed tomography (CT) and magnetic resonance imaging (MRI) scans of patients with urolithiasis. Specific details on the size and characteristics of these datasets are provided within the summary table of included studies. For instance, one study used a training set of 369 patients and a testing set of 43 patients for differentiating phleboliths from calculi. Another study analyzed CT images from 1198 urolithiasis patients for stone type prediction. No new datasets or benchmarks were introduced in this review; it synthesizes findings from existing research.

Key Insights and Ideas

  • Beyond the Naked Eye: Radiomics allows for the extraction of a vast number of quantitative features from medical images that are imperceptible to the human eye, offering a deeper, data-driven understanding of stone characteristics.
  • Phlebolith vs. Stone Differentiation: Radiomics shows significant promise in accurately distinguishing between kidney stones and phleboliths, which can reduce unnecessary interventions and associated costs.
  • Predicting Stone Composition: Pre-operative identification of stone type (e.g., calcium oxalate monohydrate) using radiomics can inform treatment strategies and potentially improve outcomes.
  • Guiding Endourological Procedures: Radiomics can predict the likelihood of spontaneous stone passage and help estimate stone burden, aiding in the selection of appropriate surgical interventions like RIRS or PCNL and predicting stone-free rates.
  • The Dawn of Personalized Urology: By providing objective, quantitative data, radiomics has the potential to move urology towards more personalized treatment plans, tailored to individual patient and stone characteristics.

Key Results

The review identified seven studies that demonstrated the utility of radiomics in urolithiasis. Key findings include:

  • Phlebolith Differentiation: One study achieved an accuracy of 85.1% and a ROC-AUC of 0.902 in differentiating kidney stones from phleboliths using a radiomics-reinforced machine-learning algorithm.
  • Infection Stone Prediction: Radiomics models have shown high accuracy (e.g., 90.7% accuracy, 0.97 AUC) in predicting infection stones, outperforming traditional clinical parameters like urine pH.
  • Stone Type Classification: Studies reported accuracies ranging from 87% to 90.7% in predicting various stone types (e.g., calcium oxalate monohydrate, struvite) using radiomics and deep learning models.
  • Predicting Stone-Free Rates: Radiomics nomograms have demonstrated strong predictive capabilities for stone-free rates after procedures like RIRS, with AUCs as high as 0.949.

Limitations and Future Work

The authors acknowledge that radiomics in urolithiasis is still in its infancy. Key limitations and future directions include:

  • Reproducibility and Transferability: Radiomics features can be sensitive to variations in imaging protocols, equipment, and processing, posing challenges for standardization and widespread adoption. The establishment of initiatives like the Image Biomarker Standardization Initiative (IBSI) is crucial for addressing this.
  • Data Dependency: Deep learning-based radiomics relies heavily on large datasets, and the features extracted are dataset-dependent. Transfer learning and the development of robust, generalizable models are needed.
  • Limited Scope: The current review found only seven studies, indicating a need for more research to explore the full potential of radiomics.
  • Future Frontiers: Future research should focus on expanding radiomics applications to extracorporeal shock wave lithotripsy (ESWL), predicting outcomes for different stone compositions, and integrating radiomics with other data sources (clinical, molecular, genomic) for a more comprehensive understanding and optimized management of urolithiasis. The ultimate goal is to develop robust, validated models that can be seamlessly integrated into clinical workflows.

Summary updated

Abstract

Radiomics is increasingly applied to the diagnosis, management, and outcome prediction of various urological conditions. Urolithiasis is a common benign condition with a high incidence and recurrence rate. The purpose of this scoping review is to evaluate the current evidence of the application of radiomics in urolithiasis, especially its utility in diagnostics and therapeutics. An electronic literature search on radiomics in the setting of urolithiasis was conducted on PubMed, EMBASE, and Scopus from inception to 21 March 2022. A total of 7 studies were included. Radiomics has been successfully applied in the field of urolithiasis to differentiate phleboliths from calculi and classify stone types and composition pre-operatively. More importantly, it has also been utilized to predict outcomes and complications after endourological procedures. Although radiomics in urolithiasis is still in its infancy, it has the potential for large-scale implementation. Its greatest potential lies in the correlation with conventional established diagnostic and therapeutic factors.
BibTeX
@article{lim2022radiomics,
  author = {Ee Jean Lim and Daniele Castellani and Wei Zheng So and Khi Yung Fong and Jing Qiu Li and Ho Yee Tiong and Nariman Gadzhiev and Chin Tiong Heng and Jeremy Yuen‐Chun Teoh and Nithesh Naik and Khurshid R. Ghani and Kemal Sarıca and Jean de la Rosette and Bhaskar Somani and Vineet Gauhar},
  title = {{Radiomics in Urolithiasis: Systematic Review of Current Applications, Limitations, and Future Directions}},
  journal = {Journal of Clinical Medicine},
  year = {2022},
  doi = {10.3390/jcm11175151},
}
Figure