CLEAR item#5

“Rationale for using a radiomic approach. Describe why a radiomics approach is considered. Performance and problematic aspects of currently used methods need to be described. Mention what the radiomics approach would offer to solve these problems. Clearly state how radiomics could affect clinical practice considering the study question.” [1] (from the article by Kocak et al.; licensed under CC BY 4.0)

Reporting examples for CLEAR item#5

Example#1. “Postoperative histopathological examination is the gold standard for determining Lauren classification, but there is a lag in obtaining Lauren classification through postoperative pathology. Although Lauren classification can be obtained preoperatively by gastroscopic biopsy, it is not only invasive but also few tissue specimens, which has a significant impact on the diagnostic accuracy of Lauren classification. The literature reports that the concordance rate of Lauren classification between biopsy and surgical samples is only 64.7%. Therefore, accurate preoperative Lauren classification of gastric cancer can facilitate individualized treatment and improve prognosis.

Computed tomography (CT) is a convenient and fast examination option for patients suspected of having gastric cancer. Some studies have found that morphological features such as tumor size, location, and enhancement pattern in gastric cancer are correlated with the Lauren classification. However, due to a lack of quantitative parameters and diagnostic thresholds, the value of traditional imaging features in predicting the Lauren classification is limited. In this regard, dual-energy CT (DECT) is a novel imaging modality that brings CT-based diagnosis from the morphological to the functional field by employing iodine mapping (IM) to quantitatively reflect the lesion’s blood supply. A few imaging studies have reported positively on the diagnostic value of iodine mapping in gastric cancer.

Radiomics transform visual information from imaging data into a large number of deep digital features for quantitative studies. Through feature extraction and dimensionality reduction, high-dimensional features with great stability and reproducibility related to the disease’s biological behavior can be used for model building, which allows improved objective quantitative assessment and has potential advantages in tumor precision assessment.” [2] (from the article by Li et al.; licensed under CC BY 4.0)

Example#2. “Despite becoming a gold standard for serial MRI assessment and demonstrating a high negative predictive value ranging between 92 and 100%, PRECISE has been exclusively validated in expert centres and still showed only a moderate positive predictive value of 15 to 66% for predicting histopathological PCa progression on AS. In parallel with working on further iterations of PRECISE, developing quantitative tools to make serial MRI assessment more objective may help further improve its performance and limit variance to achieve consistent expert-level quality. Pilot studies have adopted artificial intelligence (AI) techniques to devise MRI-derived radiomics models for predicting PCa progression on AS both at baseline and at follow-up. However, in the multiple time point follow-up setting, the only established methodological framework for analyzing temporal radiomic patterns is delta-radiomics (DR), which only measures change between two time points. Being more appropriate for treatment response assessment where clinical decisions are often based on a single post-treatment study, it is less applicable to the AS setting where patients undergo multiple scans, each storing quantitative data describing patient-specific dynamics of tumor development. In this study, we aimed to develop a time series radiomics (TSR) framework for predicting histopathological PCa progression on AS based on longitudinal changes in radiomic features extracted from all MRI scans obtained over the follow-up period. We hypothesized that the TSR predictive model incorporating all imaging and clinical data collected during AS would outperform DR and achieve at least comparable performance to the expert-derived PRECISE scoring as a clinical standard, offering a novel approach towards quantitative serial medical imaging data analysis.” [3] (from the article by Sushentsev et al.; licensed under CC BY 4.0)

Explanation and elaboration of CLEAR item#5

A valid scientific purpose must be identified before designing well-founded clinical AI studies [4]. Radiomics research is no exception, and it should be focused on developing solutions to unmet and relevant clinical needs, with the model’s potential clinical usefulness being a top priority. Nonetheless, despite the fact that the feasibility of this powerful post-processing technique has been demonstrated in a variety of settings, it appears that many use cases are proposed merely because radiomics can work, with little regard for why it would be required [5]. Can the new model fill a gap in patient management? Is it foreseeable that it would improve the patient’s outcome? Could radiomics enhance or replace current diagnostic/prognostic standards? These types of questions must be addressed in the first place to justify and embrace the complexity that characterizes radiomics.

References

  1. Kocak B, Baessler B, Bakas S, et al (2023) CheckList for EvaluAtion of Radiomics research (CLEAR): a step-by-step reporting guideline for authors and reviewers endorsed by ESR and EuSoMII. Insights Imaging 14:75. https://doi.org/10.1186/s13244-023-01415-8
  2. Li M, Qin H, Yu X, et al (2023) Preoperative prediction of Lauren classification in gastric cancer: a radiomics model based on dual-energy CT iodine map. Insights Imaging 14:125. https://doi.org/10.1186/s13244-023-01477-8
  3. Sushentsev N, Rundo L, Abrego L, et al (2023) Time series radiomics for the prediction of prostate cancer progression in patients on active surveillance. Eur Radiol 33:3792–3800. https://doi.org/10.1007/s00330-023-09438-x
  4. Koçak B, Cuocolo R, dos Santos DP, et al (2023) Must-have Qualities of Clinical Research on Artificial Intelligence and Machine Learning. Balk Med J 40:3–12. https://doi.org/10.4274/balkanmedj.galenos.2022.2022-11-51
  5. Stanzione A (2022) Feasible does not mean useful: Do we always need radiomics? Eur J Radiol 156:110545. https://doi.org/10.1016/j.ejrad.2022.110545

Back Next