Using 2D U-Net convolutional neural networks for automatic acetabular and proximal femur segmentation of hip MRI images and morphological quantification: a preliminary study in DDH

Table 3 The groups of hip MRI images and the information about age and slice numbers of each group

Groups	Serial number combination	Age corresponding to serial number	Number of slices in validation data	Number of slices in training data
Group-1	1 + 15 + 26	12, 1, 7	124	1260
Group-2	2 + 3 + 29	6, 3, 11	121	1263
Group-3	4 + 12 + 24	1, 8, 12	129	1255
Group-4	5 + 25 + 28	4, 10, 5	126	1258
Group-5	6 + 16 + 30	1, 11, 8	122	1262
Group-6	7 + 13 + 14	9, 2, 8	122	1262
Group-7	8 + 11 + 23	6, 10, 4	128	1256
Group-8	9 + 10 + 32	9, 3, 8	129	1255
Group-9	17 + 19 + 22	5, 9, 5	125	1259
Group-10	18 + 27 + 33	7, 8, 2	121	1263
Group-11	20 + 21 + 31	7, 13, 3	137	1247

MRI slices from 33 children were used with Leave-One-Out Cross-Validation (LOOCV). Data were divided into 11 groups, training 11 models, with 10 groups for training and 1 for validation to maximize sample utilization

ISSN: 1475-925X