18 June 2020 : Clinical Research
Ensemble Deep Learning Model for Multicenter Classification of Thyroid Nodules on Ultrasound Images
Xi Wei1ACDEG*, Ming Gao2BEF, Ruiguo Yu3BCD, Zhiqiang Liu3BCE, Qing Gu4BC, Xun Liu5BC, Zhiming Zheng6BC, Xiangqian Zheng2BC, Jialin Zhu1ABCDEF, Sheng Zhang1BDDOI: 10.12659/MSM.926096
Med Sci Monit 2020; 26:e926096
Table 2 Demographic data and image information for all patients from four medical centers.
| Center 1 | Center 2 | Center 3 | Center 4 | |
|---|---|---|---|---|
| 10993 | 151 | 460 | 261 | |
| Female (n) | 8379 (76.22%) | 117 (77.48%) | 369 (80.22%) | 197 (75.48%) |
| Male (n) | 2614 (23.78%) | 34 (22.52%) | 91 (19.78%) | 64 (24.52%) |
| 46 (18–84) | 49 (21–67) | 51 (18–73) | 52 (23–70) | |
| Left lobe | 5063 (46.06%) | 79 (52.31%) | 214 (46.52%) | 127 (48.66%) |
| Right lobe | 5652 (51.41) | 70 (46.36%) | 228 (49.57%) | 128 (27.77%) |
| Isthmus | 278 (2.53%) | 2 (1.32%) | 18 (3.91%) | 6 (2.61%) |
| 1.25 (0.38–7.80) | 1.72 (0.58–6.25) | 1.53 (0.30–6.91) | 2.12 (0.49–7.21) | |
| 3996 (36.35%) | 82 (54.30%) | 213 (46.30%) | 153 (58.62%) | |
| Nodular goiter | 2745 | 71 | 166 | 127 |
| Adenomatous goiter | 551 | 11 | 45 | 26 |
| Thyroid granuloma | 518 | |||
| Follicular adenoma | 182 | 2 | ||
| 6997 (63.65%) | 69 (45.70%) | 247 (53.70%) | 108 (41.38%) | |
| PTC | 6910 | 69 | 246 | 108 |
| MTC | 65 | 1 | ||
| FTC | 20 | |||
| ATC | 2 | |||
| 23504 | 530 | 1475 | 1032 | |
| Benign | 9084 | 289 | 881 | 502 |
| Malignant | 14420 | 241 | 594 | 530 |
| Philips EPIQ 5 | Philips IU22 | Philips EPIQ 7 | GE LOGIQ E9 | |
| Philips IU Elite | Philips IU22 | Mindray DC-8 | ||
| Philips IU22 | Siemens Acuson Qxana 2 | Siemens Acuson S2000 | ||
| Philips HD11 | Siemens Acuson S2000 | |||
| TOSHIBA Aplio 500 | ||||
| TOSHIBA Aplio 400 | ||||
| PTC – papillary thyroid carcinoma; MTC – medullary thyroid carcinoma; FTC – follicular thyroid carcinoma; ATC – anaplastic thyroid carcinoma. | ||||