Table 3 Indicator, algorithm, and data source for the included studies
First author and year | Indicator definition | Algorithm | Data source | ||||||
|---|---|---|---|---|---|---|---|---|---|
Method for predictor | Exclusion of poor quality imaging | Heatmap provided | Algorithm architecture | Transfer learning applied | Source of data | Number of images for training/internal/external | Data range | Open access data | |
Zhou Z, 2021 | PET/CT images | N/A | N/A | DLCNN | Yes | Retrospective, data from 33 centers. | 110/110/32 | 2007/05–2018/10 | No |
McAvoy M, 2021 | MRI | Yes | N/A | CNN, EfficientNet | Yes | Retrospective, data from the Partners Healthcare Research Patient Data Registry (RPDR) web-based query tool. | 1245/202/NR | 2015–2018 | No |
Li D, 2020 | Histopathology images | N/A | N/A | GOTDP-MP-CNNs | Yes | Retrospective, data from 3 hospitals. | 80%/10%/NR | NR | Yes |
Miyoshi H, 2020 | WSI | N/A | N/A | DCNN | No | Retrospective, data from Kurume University. | 388/NR/NR | 2010–2017 | NR |
Park JE, 2020 | Ultrasound | N/A | N/A | CART | No | Retrospective, data from a single center in US. | 170/NR/72 | 2012–2018 | NR |
Mohlman JS, 2020 | Histologic images | Yes | N/A | Deep dense CNN | Yes | Retrospective, data from hematopathology archives at author’s institutions. | 8796/NR/NR | 2010–2020 | NR |
Achi HE, 2019 | WSI | N/A | N/A | DLCNN | No | Retrospective, data from Virtual Pathology at the University of Leeds and Virtual Slide Box from University of Iowa. | 1856/464/NR | NR | NR |
Im H, 2018 | Holograms | N/A | N/A | CNN | No | Prospective. | 3447/1723/NR | NR | Yes |
Guan Q, 2019 | Cytological images | No | No | DCNN, Inception-v3 | Yes | Retrospective, data from a single center. | 156/26/NR | 2016.11–2017.11 | NR |
Guo R, 2021 | PET/CT images | No | No | ResNet-18 | No | Retrospective and prospective, data from Shanghai Ruijin Hospital. | 64/20/NR | 2011/06–2020/10 | Yes |
Xia W, 2021 | MRI | N/A | N/A | IF-CNN | Yes | Retrospective, data from a single center. | 80%/20%/NR | 2011–2019 | NR |
Zhang Y, 2021 | MRI | Yes | No | 3D U-Net/Resnet18 | No | Retrospective, data from two centers in China. | 65/27/NR | 2005.1–2019.12 | Yes |
Syrykh C, 2020 | WSI | N/A | N/A | Stochastic BNN | No | Retrospective, data from the lymphopath database in Toulouse University Cancer Institute and Dijon University Hospital, France. | 50%/25%/25% | NR | No |
Wang H, 2020 | PET/CT images | No | No | LASSO | No | Retrospective. | 82/28/NR | 2011/01–2017/01 | NR |
Zhang J, 2020 | Histopathology images | No | N/A | VGG-16, VGG-19, ResNet-50, DenseNet-121 | Yes | Retrospective, data from a reference dataset provided by Janowczyk et al. | 269/30/NR | NR | NR |
Wang Q, 2017 | Microscopic blood images | No | No | ANN | No | Retrospective, data from the Department of Hematology, Ruijin Hospital, Shanghai, China. | 27/NR/NR | NR | NR |
Schouten JPE, 2021 | Cytological images | No | No | CNN | No | Retrospective, data from the Department of Information Technology at the University of Milan. | 200/25/25 | NR | NR |
Nakagawa M, 2018 | MRI | N/A | N/A | Machine learning classifiers | No | Retrospective. | 63/7/NR | 2008.01–2013.12 | NR |
Shafique S, 2018 | Microscopic blood images | No | No | DCNN | Yes | Retrospective, data from ALL-Image DataBase (IDB). | 60%/40%/NR | NR | Yes |
Kong Z, 2019 | PET/CT images | N/A | Yes | Decision tree classifier | No | Retrospective, data from Peking Union Medical College Hospital. | 80%/20%/NR | 2010/01–2018/10 | NR |
Weisman AJ, 2020 | PET/CT images | No | No | CNN | No | Retrospective, data from one of two separate prospective multicenter imaging trials. | 65%/15%/20% | 2005–2011 | NR |
Kim Y, 2018 | mpMRI | Yes | No | Logistic regression-based classifier, random forest | No | Retrospective, data from two independent cohorts: the discovery cohort from Samsung Medical Center in Seoul, South Korea; and the validation cohort from Asan Medical Center in Seoul, South Korea. | 36/36/28 | 2013/01–2016/02 | NR |
Styczeń M, 2012 | Specimens | No | No | Neural network | No | Retrospective, data from the Department of Pathology. | NR/NR/NR | 2005–2009 | NR |
Guo J, 2018 | MRI | Yes | No | LASSO | No | Retrospective, data from a single center. | 47/37/NR | 2010/03–2016/07 | NR |
Azamossadat H, 2023 | Microscopic blood images | No | No | CNN | Yes | Restrospective, data from several Tehran hospitals | 85/5/10 | NR | NR |
Chava P, 2023 | Biopsy slides | No | No | MIL | No | Restrospective, data from the Tel Aviv Sourasky Medical Center | 55/NR/25 | 2017/01–2022/01 | NR |
Jermphiphut J, 2023 | MRI | No | No | CNN | No | Restrospective, data from a historical cohort study | 709/177/NR | 2010/01–2021/12 | NR |
Hikaru A, 2023 | PET/CT images | No | Yes | CNN | No | Restrospective, data from Nthe Tokyo Medical and Dental University, Tokyo, Japan, and Shinjuku Tsurukame Clinic, Tokyo, Japan. | 62/NR/NR | 2005/08–06/2020 | NR |
Manjit K, 2023 | Peripheral blood smear images | No | No | DSCNet | No | Restrospective, data from Acute Lymphoblastic Leukemia Image Dataset | 20,000/NR/NR | 2021 | Yes |
Noriaki H, 2023 | WSI | No | No | ResNet50 | No | Restrospective, data from Kurume University and Nagoya University Hospital | 249/NR/208 | 2018 | NR |