Deep learning with nuke med quantifies prostate gland uptake, predicts cancer survival
European researchers have trained and validated a deep-learning algorithm for quantifying prostate measurements on PET/CT scans. They also showed that the measurements can serve as biomarkers useful in formulating clinical prognoses for prostate cancer patients.
Eirini Polymeri, MD, of the University of Gothenburg in Sweden and colleagues had their study published online Dec. 3 in Clinical Physiology and Functional Imaging.
The team trained the algorithm on prostate volumes in manually segmented CT images from 100 patients. They validated it on 45 patients who had biopsy-proven prostate cancer that had not been treated with hormonal or androgen deprivation therapy.
The researchers compared the algorithmic measurements of prostate volume to manual measurements made independently by two physicians. They further analyzed associations between the automated PET/CT biomarkers and age, prostate specific-antigen (PSA) and Gleason score, evaluating overall survival using a univariate regression model.
Checking agreement between the algorithm and the observers using the Sørensen‐Dice index (SDI), which gauges the similarity of two samples, the authors found the SDI between the automated and manual volume segmentations was, respectively, 0.78 and 0.79.
Meanwhile, Polymeri et al. reported, “PET/CT quantifications of total lesion uptake and the relation between volume of abnormal voxels and total prostate volume were significantly associated with overall survival whereas age, PSA and Gleason score were not.”
Dave P. has worked in journalism, marketing and public relations for more than 30 years, frequently concentrating on hospitals, healthcare technology and Catholic communications. He has also specialized in fundraising communications, ghostwriting for CEOs of local, national and global charities, nonprofits and foundations.