Machine learning-based prediction of invisible intraprostatic prostate cancer lesions on 68 Ga-PSMA-11 PET/CT in patients with primary prostate cancer

Abstract

Purpose 68 Ga-PSMA PET/CT has high speci city and sensitivity for the detection of both intraprostatic tumor focal lesions and metastasis. However, approximately 10% of primary prostate cancer are invisible on PSMA-PET (exhibit no or minimal uptake). In this work, we investigated whether machine learningbased radiomics models derived from PSMA-PET images could predict invisible intraprostatic lesions on 68 Ga-PSMA-11 PET in patients with primary prostate cancer.Methods In this retrospective study, patients with or without prostate cancer who underwent 68 Ga-PSMA PET/CT and presented negative on PSMA-PET image at either of two different institutions were included: institution 1 (between 2017 to 2020) for the training set and institution 2 (between 2019 to 2020) for the external test set. Three random forest (RF) models were built using selected features extract from standard PET images, delayed PET images, and both standard and delayed PET images. Then, subsequent 10-fold cross-validation was performed. In the test phase, the three RF models and PSA density (PSAD, cut-off value: 0.15ng/ml/ml) were tested with the external test set. The area under the receiver operating characteristic curve (AUC) was calculated for the models and PSAD. The AUCs of the radiomics model and PSAD were compared.Results A total of 64 patients (39 with prostate cancer and 25 with benign prostate disease) were in the training set, and 36 (21 with prostate cancer and 15 with benign prostate disease) were in the test set.The average AUCs of the three RF models from 10-fold cross-validation were 0.87 (95% CI: 0.72, 1.00), 0.86 (95% CI: 0.63, 1.00) and 0.91 (95% CI: 0.69, 1.00), respectively. In the test set, the AUCs of the three trained RF models and PSAD were 0.903 (95% CI: 0.830, 0.975), 0.856 (95% CI: 0.748, 0.964), 0.925 (95% CI:0.838, 1.00), and 0.662 (95% CI: 0.510, 0.813). The AUCs of the three radiomics models were higher than that of PSAD (0.903, 0.856 and 0.925 vs 0.662, respectively; P = .007, P = .045 and P = .005, respectively).Conclusion Random forest models developed by 68 Ga-PSMA-11 PET-based radiomics features were proven useful for accurate prediction of invisible intraprostatic lesion on 68 Ga-PSMA-11 PET in patients with primary prostate cancer and showed better diagnostic performance compared with PSAD.

Predicting clinically significant prostate cancer with a deep learning approach: a multicentre retrospective study

“…This study found that the performances of these RF models were better than that of the PSA density (PSAD). The AUCs (0.856-0.925) of the RF models in Yi et al [35] were also slightly higher than those of the DL models in the present study. Although the number of samples in this study (training set: 64, testing set: 36) was much lower than those in the present study (training set: 1216, testing set: 645), this may demonstrate the potential of multimodal images to aid in the precise diagnosis of PCa.…”

“…Although the NCCN guidelines recommend MRI as the first and most important method for monitoring PCa, some studies have used PET for the diagnosis of PCa (e.g. [35][36][37][38]). Among these examples, Yi et al [35] developed three random forest (RF) models based on PET images from two centres for the classification between PCa and non-PCa.…”

“…[35][36][37][38]). Among these examples, Yi et al [35] developed three random forest (RF) models based on PET images from two centres for the classification between PCa and non-PCa. This study found that the performances of these RF models were better than that of the PSA density (PSAD).…”

A review of artificial intelligence in prostate cancer detection on imaging

“…One such modality is Gallium-68 PSMA-11 PET-CT scans. 204 Recent studies demonstrate that PSMA PET-CT scans can significantly improve prostate cancer detection and treatment planning, are more accurate than conventional imaging with CT and bone scanning, 205 add value to MRI for diagnosis, 206 and may enable better prediction of pre-operative pathological outcomes than MRI.207–209 There is an opportunity for AI to address shortcomings of these new imaging modalities, but due to its recent FDA approval, only a few published AI models 32 , 210 , 211 exist for prostate cancer detection using PSMA PET currently.…”

Radiomics in Oncological PET Imaging: A Systematic Review—Part 2, Infradiaphragmatic Cancers, Blood Malignancies, Melanoma and Musculoskeletal Cancers

“…We included 11 studies of prostate cancer radiomics [ 102 , 103 , 104 , 105 , 106 , 107 , 108 , 109 , 110 , 111 , 112 ], 8 employing 68Ga-PSMA, 2 Choline (1 18F-Fluoroethilcholine and 1 11C-Choline) and 1 using 18F-DCFPyl. An average of 71.3 patients was included (range 41–101); 2 of the studies were prospective and 4/11 studies used a validation cohort.…”