Signal-based AI-driven software solution for automated quantification of metastatic bone disease and treatment response assessment using Whole-Body Diffusion-Weighted MRI (WB-DWI) biomarkers in Advanced Prostate Cancer (2505.09011v1)

Abstract: We developed an AI-driven software solution to quantify metastatic bone disease from WB-DWI scans. Core technologies include: (i) a weakly-supervised Residual U-Net model generating a skeleton probability map to isolate bone; (ii) a statistical framework for WB-DWI intensity normalisation, obtaining a signal-normalised b=900s/mm^2 (b900) image; and (iii) a shallow convolutional neural network that processes outputs from (i) and (ii) to generate a mask of suspected bone lesions, characterised by higher b900 signal intensity due to restricted water diffusion. This mask is applied to the gADC map to extract TDV and gADC statistics. We tested the tool using expert-defined metastatic bone disease delineations on 66 datasets, assessed repeatability of imaging biomarkers (N=10), and compared software-based response assessment with a construct reference standard based on clinical, laboratory and imaging assessments (N=118). Dice score between manual and automated delineations was 0.6 for lesions within pelvis and spine, with an average surface distance of 2mm. Relative differences for log-transformed TDV (log-TDV) and median gADC were below 9% and 5%, respectively. Repeatability analysis showed coefficients of variation of 4.57% for log-TDV and 3.54% for median gADC, with intraclass correlation coefficients above 0.9. The software achieved 80.5% accuracy, 84.3% sensitivity, and 85.7% specificity in assessing response to treatment compared to the construct reference standard. Computation time generating a mask averaged 90 seconds per scan. Our software enables reproducible TDV and gADC quantification from WB-DWI scans for monitoring metastatic bone disease response, thus providing potentially useful measurements for clinical decision-making in APC patients.

Automated Quantification of Metastatic Bone Disease Using AI-Driven WB-DWI Analysis in Advanced Prostate Cancer

The presented paper explores the integration of artificial intelligence with Whole-Body Diffusion-Weighted MRI (WB-DWI) for the automated assessment of metastatic bone disease in patients suffering from Advanced Prostate Cancer (APC). The proposed software solution aims to address the clinical necessity for rapid and reproducible quantification of treatment response using WB-DWI biomarkers, such as Total Diffusion Volume (TDV) and global Apparent Diffusion Coefficient (gADC).

Utilizing a combination of AI-driven methodologies, the authors have developed a workflow that involves several key components: weakly-supervised Residual U-Net models to generate skeleton probability maps, intensity normalization techniques to produce signal-normalized b900 images, and shallow convolutional neural networks for masked delineation of bone lesions. These steps culminate in the extraction of TDV and gADC statistics meant to facilitate the monitoring of disease spread and treatment efficacy.

Key findings in this study include the achievement of a Dice score of 0.6, indicating fair agreement between manual and automated delineations within critical areas like the pelvis and spine. Notably, the software displayed strong repeatability characterized by coefficients of variation for log-transformed TDV and median gADC at 4.57% and 3.54%, respectively, with intraclass correlation coefficients exceeding 0.9. In comparison to a construct reference method combining clinical, laboratory, and imaging evaluations, the software demonstrated an accuracy of 80.5%, sensitivity of 84.3%, and specificity of 85.7% in appraising treatment response.

The implications of this research are manifold. Practically, the automated solution promises to streamline the diagnostic workflow in oncology, enabling timely and objective evaluation of metastatic progression and therapeutic impacts without excessive reader variability common in manual assessments. Theoretically, this research may act as a springboard for further machine learning advancements in medical imaging, enhancing precision in staging and prognosis of APC.

While the results are promising, limitations include potential generalizability concerns given the homogeneous imaging datasets from a single MRI vendor and algorithmic reliance on signal-based rather than multi-parametric inputs. Future iterations of the software could benefit from incorporating relative fat fraction measurements and expanding analyses to include soft tissue metastases.

In conclusion, the confluence of AI with advanced imaging techniques presents an encouraging avenue for oncological care, specifically in automated response assessments. A prospective multi-center clinical trial is underway to substantiate the software's clinical utility, which may ultimately redefine the diagnostic and treatment landscape for APC patients.