Table II.

Overview of the studies on AI in WSI.

A, Prediction of different pathological subtypes
First author, year	Disease	Number of patients	AI model	Main result	Main conclusion	(Refs.)
Farahani et al, 2022	OC	545	ML	The best-performing model achieved a diagnostic concordance of 81.38% (Cohen's κ, 0.7378) in the training set and 80.97% concordance (Cohen's κ, 0.7547) in the external dataset.	The CNN model may improve the diagnostic efficiency for determining OC pathological subtypes.	(96)
Wang et al, 2022	OC	288	DL	For an independent testing set, the three proposed methods obtained promising results with high recall (sensitivity) values of 0.946, 0.893 and 0.964, respectively.	The DL method can help identify patients with different treatment responses.	(97)
B, Predict the mutation status of a gene
First author, year	Disease	Number of patients	AI model	Main result	Main conclusion	(Refs.)
Ho et al, 2023	OC	609	DL	The model achieved an intersection-over-union value of 0.74, a recall value of 0.86 and a precision value of 0.84.	The DL model can be used to diagnose OC and find novel morphological patterns to predict molecular subtypes.	(98)
Nero et al, 2022	HGSOC	644	DL	The model achieved an AUC of 0.71, with a negative predictive value of 0.69 and a positive predictive value of 0.75 when applied to predict PFS.	The DL model based on WSI can predict BRCA1/2 gene status.	(99)
C, Predict the efficacy and prognosis of drug therapy
First author, year	Disease	Number of patients	AI model	Main result	Main conclusion	(Refs.)
Laury et al, 2021	HGSOC	30	CNN	The CNN model based on WSI discriminated the response to primary platinum-based chemotherapy with high sensitivity (73%) and specificity (91%).	DL based image analysis is able to. predict outcome	(100)
Wang et al, 2022	EOC	720	DL	The model in combination with AIM2 achieves high accuracy (0.92), recall (0.97), F-measure (0.93) and AUC (0.97) values for the first experiment (66% training and 34% testing) and high accuracy (0.86±0.07), precision (0.9±0.07), recall (0.85±0.06), F-measure (0.87±0.06) and AUC (0.91±0.05) for the second experiment using five-fold cross validation, respectively.	AIM2-DL model can distinguish patients gaining positive therapeutic effects with low cancer recurrence from patients with disease progression after treatment.	(102)
Wu et al, 2022	OC	90	DL	The mean value of the resulting C-index was 0.5789 (range, 0.5096-0.6053), and the resulting P-value was 0.00845.	The DL framework is a promising method for searching WSIs and providing a valuable clinical means for prognosis.	(103)

AI, artificial intelligence; OC, ovarian cancer; EOC, epithelial ovarian cancer; HGSOC, high-grade serous ovarian cancer; ML, machine learning; DL, deep learning; CNN, convolutional neural network; AUC, area under the curve; PFS, progression-free survival; WSI, whole slide imaging.