AI-powered MRI predicts prostate cancer outcomes

AI-powered MRI analysis provides new insights into prostate cancer prognosis and accurately predicts prostate cancer prognosis metastasis Risks and treatment outcomes to improve patient care.

study: AI-derived tumor volume and localized prostate cancer outcome from multiparametric MRI. Image credit: Shutterstock AI / Shutterstock.com

In a recent study published in RadiologyResearchers use artificial intelligence (AI)-based magnetic resonance imaging (MRI) data to measure tumor volume inside the prostate to determine outcomes, including metastatic risk, for treated prostate cancer patients. We are determining whether it can be predicted. Radiology Or surgery.

Advances in MRI are changing prostate cancer detection and diagnosis

Multiparametric MRI combines multiple MRI techniques to create detailed images of internal anatomy. This imaging technology has transformed the management of prostate cancer by improving the detection of serious cases while minimizing the detection rate of non-essential disease. MRI-guided biopsies also significantly improve the accuracy of cancer diagnosis.

MRI shows various characteristics of prostate cancer. These include the Prostate Imaging Reporting and Data System (PI-RADS) score, the lesion score, and the radiological T stage, which indicates the extent of the tumor's spread within the prostate. . Analyzing these characteristics can indicate the potential recurrence rate of prostate cancer. However, the evaluation of these features varies between observers. Different tumor grading systems are associated with different accuracies, which further complicates diagnostic consistency.

The use of AI has the potential to increase the clinical value of MRI by providing consistent analysis of images. Recent studies on deep learning models have shown accuracy levels comparable to experienced radiologists for delineating tumors within the prostate.

About research

Current study determines whether calculating tumor volume using an AI-based approach can provide independent prognostic insight for prostate cancer patients who have previously undergone surgery or radiotherapy It was intended for that purpose. These results were compared with those obtained from standard MRI evaluation.

This retrospective study included prostate cancer patients who underwent MRI scans before undergoing radical prostatectomy or radiation therapy. Patient data were collected from medical records and consisted of clinical, pathological, and treatment information, including tumor classification based on PI-RADS and National Comprehensive Cancer Network (NCCN) scores.

The biochemical failure is an increase in prostate-specific cells. antigen (PSA) levels after treatment such as radical prostatectomy or radiation therapy. In the current study, biochemical failure was defined as an increase in PSA concentration of at least 2 ng/mL above the nadir level after treatment with radiotherapy, and clinical progression or PSA of at least 0.1 ng/mL in case of radical prostatectomy. was defined as an increase in .

Reference segmentations were created manually by a genitourinary radiation oncologist to outline prostate regions such as translational and peripheral regions and lesions with PI-RADS scores of 3–5.

The AI ​​model nnU-Net, a deep learning-based segmentation method, was trained to delineate prostate regions and tumors from various MRI sequences. The model was then validated using a subgroup of images from patients who received radiotherapy, and then tested on images from both the radiotherapy and radical prostatectomy groups. The tumor volume was then calculated based on AI and compared with the reference volume generated by manual segmentation.

For statistical analysis, baseline comparisons between cross-validation and study radiotherapy groups were performed by Wilcoxon rank sum test and Fisher's exact test for continuous variables and categorical data, respectively. Sensitivity and positive predictive value were used to evaluate the accuracy of the AI ​​model in tumor detection.

Research results

Total intraprostatic tumor volume (V_A.I.) was a strong independent predictor of outcome in patients with localized prostate cancer who underwent radiotherapy or radical prostatectomy. In fact, volume predicted by AI was significantly associated with metastasis and biochemical failure.

For the radiotherapy group, V_A.I. The prediction accuracy for 7-year metastasis was higher compared to traditional risk groups. Furthermore, the predictive information provided by V is_A.I. was comparable to intraprostatic tumor volume by manual reference segmentation, demonstrating consistency of results and reliability as a tool to predict patient outcome. The AI ​​algorithm sometimes missed lesions with a PI-RAD score of 5, but_A.I. remained sensitive to clinically significant disease burden.

Ability of nnU-Net to predict metastasis using V_A.I. were comparable to or better than novel genomic or computational pathology biomarkers. Therefore, this AI tool has the potential to improve treatment planning by identifying patients who may require a more personalized or aggressive treatment approach.

conclusion

V_A.I. This appears to be a consistent and promising approach for predicting prognosis in cases of localized prostate cancer after patients undergo radical prostatectomy or radiotherapy.

Accuracy of V_A.I. Its adaptability to a variety of imaging conditions and its robust predictive ability highlight its potential as a supplement or alternative to traditional radiological or clinical prognostic methods.