. 2023 Feb 27;14:100301. doi: 10.1016/j.jpi.2023.100301

Table 1.

Average precision for deployment vs. fine-tuning for n = 1, 5, 10, 15, and 22 training whole slide images (mean of 5 training repetitions). The highest average precision per cell class and per tumor indication is highlighted in bold. HNSCC: head and neck squamous cell carcinoma, NSCLC: non-small cell lung cancer, TNBC: triple-negative breast cancer, GC: gastric cancer.

	Tumor cells				Non-specified cells				CD3⁺ cells
	HNSCC	NSCLC	TNBC	GC	HNSCC	NSCLC	TNBC	GC	HNSCC	NSCLC	TNBC	GC
RetinaNet₁	0.70	0.54	0.61	0.73	0.49	0.45	0.54	0.48	0.70	0.69	0.57	0.65
RetinaNet_1,T	0.70	0.58	0.74	0.83	0.49	0.45	0.63	0.55	0.70	0.69	0.72	0.69
RetinaNet₅	0.78	0.61	0.64	0.80	0.62	0.50	0.61	0.59	0.70	0.67	0.58	0.67
RetinaNet_5,T	0.78	0.62	0.76	0.83	0.62	0.55	0.66	0.63	0.70	0.69	0.73	0.70
RetinaNet₁₀	0.81	0.68	0.74	0.84	0.63	0.49	0.61	0.64	0.73	0.74	0.65	0.70
RetinaNet_10,T	0.81	0.66	0.78	0.85	0.63	0.50	0.64	0.63	0.73	0.73	0.72	0.70
RetinaNet₁₅	0.81	0.69	0.77	0.83	0.64	0.52	0.66	0.66	0.73	0.70	0.61	0.68
RetinaNet_15,T	0.81	0.68	0.75	0.84	0.64	0.51	0.65	0.65	0.73	0.73	0.71	0.72
RetinaNet₂₂	0.81	0.69	0.76	0.85	0.64	0.57	0.70	0.68	0.74	0.73	0.64	0.70
RetinaNet_22,T	0.81	0.67	0.78	0.85	0.64	0.50	0.66	0.66	0.74	0.73	0.72	0.69
Benchmark_10,T		0.67	0.78	0.85		0.54	0.68	0.69		0.73	0.72	0.71