Effectiveness of an auxiliary diagnosis system for ultrasound classification of cystic echinococcosis based on the convolution neural network: A multicenter study

doi:10.12140/j.issn.1000-7423.2025.05.004

Abstract

Abstract:

Objective To examine the value of the convolution neural network-based ResNet50 model in assisting ultrasound classification of hepatic cystic echinococcosis (HCE). Methods Ultrasound images of intrahepatic lesions were retrospectively collected from HCE patients that were definitely diagnosed with postoperative pathology or long-term follow-up of multiple imaging tools admitted to five medical institutions including First Affiliated Hospital of Xinjiang Medical University (groups A to E, among them, group A is a provincial-level tertiary grade A hospital, groups B to D are prefecture level tertiary grade A hospitals, and group E is a county-level secondary Grade A hospital; and four grassroots hospitals were merged into group F and 5 hospitals merged into group G) during the period from January 2020 through June 2024. All images were classified by a junior sonographer and a senior sonographer according to five types of CE1 (univesicular cysts), CE2 (multiple daughter cysts), CE3 (cyst with detachment of laminated membrane), CE4 (heterogeneous or hyperechoic degenerative contents), and CE5 (calcified cysts), and uploaded to the ResNet50 model-based intelligent classification diagnosis ultrasound imaging cloud platform for classification diagnosis of HCE. The diagnostic accuracy of HCE classification by junior sonographers, senior sonographers and ResNet50 model was compared among groups using chi-square test, and pairwise comparisons were conducted with Bonferroni correction. In addition, the diagnostic consistency was evaluated with kappa analysis. Results A total of 1 596 ultrasound images of HCE were included from 952 HCE patients. The diagnostic accuracy of HCE classification by junior sonographers, senior sonographers and ResNet50 model were 88.3% (594/673), 95.7% (644/673) and 89.2% (600/673) in Group A, 72.7% (157/216), 86.6% (187/216), and 89.8% (194/216) in Group B, 77.6% (232/299), 87.6% (262/299) and 91.3% (273/299) in Group C, 74.5% (164/220), 85.5% (188/220) and 90.5% (199/220) in Group D, 69.1% (130/188), 84.0% (158/188) and 88.3% (166/188) in Group E, 75.1% (693/923), 86.1% (795/923) and 90.1% (832/923) in Group F and 80.6% (1 287/1 596), 90.2% (1 439/1 596) and 89.7% (1 432/1 596) in Group G. The overall diagnostic accuracy of HCE classification was lower by the ResNet50 model diagnosis than by senior sonographers in Group A (χ² = 27.140, P < 0.05), and the overall diagnostic accuracy was higher by the ResNet50 model than by junior sonographers in groups B, C, D, and E (χ² = 25.381, 24.307, 21.121 and 24.213; all P < 0.05). The overall diagnostic accuracy of HCE classification was higher by the ResNet50 model than by junior and senior sonographers in Group F (χ² = 82.654, P < 0.05), and the overall diagnostic accuracy was higher by the ResNet50 model than by junior sonographers in Group G (χ² = 80.749, P < 0.05). Moreover, all pairwise comparisons after Bonferroni correction were statistically significant with P < 0.025. The ResNet50 model demonstrated the highest diagnostic consistency with senior sonographers for HCE classification across all groups (kappa ≥ 0.836), followed by between junior and senior sonographers (0.754 ≤ kappa ≤ 0.863), and the lowest diagnostic consistency was seen between the ResNet50 model and junior sonographers (0.674 ≤ kappa ≤ 0.855). In the ResNet50 model, CE1 was most likely to be misdiagnosed as CE3 (68.6%, 24/35), CE2 misdiagnosed as CE4 (37.9%, 11/29) or CE3 (31.0%, 9/29), CE3 misdiagnosed as CE2 (65.6%, 21/32), and CE4 and CE5 were likely to be mutually misdiagnosed (61.9%, 26/42; 57.9%, 22/38). Conclusion The convolutional neural network-based ResNet50 model is effective to decipher the ultrasound features of HCE classification, which provides auxiliary diagnostic supports to doctors in primary medical institutions, and improves the accuracy of classification interpretation.

Key words: Hepatic cystic chinococcosis, Artificial intelligence, Convolutional neural network, Ultrasound examination, Diagnostic accuracy

CLC Number:

R532.32

TIAN Jing, WANG Xiaorong, YILIYASI Abuduaini, YANG Lingfei, CHEN Lu, GAO Hongying, ZHOU Na, LI Yuling, CUI Liang, SONG Tao. Effectiveness of an auxiliary diagnosis system for ultrasound classification of cystic echinococcosis based on the convolution neural network: A multicenter study[J]. CHINESE JOURNAL OF PARASITOLOGY AND PARASITIC DISEASES, 2025, 43(5): 619-626.

Figures/Tables 7

Fig. 1

Table 1

Table 2

Table 3

Table 4

Diagnostic consistency of HCE classification by the ResNet50 model in group A-G

HCE分型（n） HCE category（n）	ResNet50模型 ResNet50 model		低年资医师 Junior physicians		高年资医师Senior physcians		χ²值 χ²value	P值 P value
HCE分型（n） HCE category（n）	病例数 No. case	准确率/% Accuracy rate/%	病例数 No. case	准确率/% Accuracy rate/%	病例数 No. case	准确率/% Accuracy rate/%	χ²值 χ²value	P值 P value
A组 A group
CE1（176）	157	89.2	155	88.1	167	94.9	5.579	0.061
CE2（181）	167	92.3	170	93.9	176	97.2	4.446	0.108
CE3（81）	69	85.2	65	80.2	77	95.1	8.062	0.018
CE4（150）	132	88.0	130	86.7	142	94.7	6.005	0.049
CE5（85）	75	88.2	74	87.1	82	96.5	5.244	0.073
合计 Total（673）	600	89.2	594	88.3	644	95.7^a	27.140	< 0.001
B组 B group
CE1（49）	45	91.8	38	77.6	44	89.8	4.977	0.083
CE2（44）	40	90.9	33	75.0	38	86.4	4.417	0.110
CE3（39）	33	84.6	25	64.1	34	87.2	7.427	0.024
CE4（50）	45	90.0	37	74.0	42	84.0	4.560	0.102
CE5（34）	31	91.2	24	70.6	29	85.3	5.262	0.072
合计 Total（216）	194	89.8	157	72.7^b	187	86.6	25.381	< 0.001
C组 C group
CE1（45）	41	91.1	34	75.6	39	86.7	4.399	0.111
CE2（47）	44	93.6	37	78.7	41	87.2	4.501	0.105
CE3（36）	31	86.1	23	63.9	30	83.3	6.107	0.047
CE4（120）	111	92.5	100	83.3	107	89.2	5.013	0.082
CE5（51）	46	90.2	38	74.5	45	88.2	5.634	0.060
合计 Total（299）	273	91.3	232	77.6^b	262	87.6	24.307	< 0.001
D组 D group
CE1（43）	39	90.7	32	74.4	37	86.0	4.437	0.109
CE2（46）	42	91.3	35	76.1	40	87.0	4.381	0.112
CE3（38）	34	89.5	28	73.7	32	84.2	3.396	0.183
CE4（52）	47	90.4	39	75.0	44	84.6	4.523	0.104
CE5（41）	37	90.2	30	73.2	35	85.4	4.479	0.107
合计 Total（220）	199	90.5	164	74.5^b	188	85.5	21.121	< 0.001
E组 E group
CE1（40）	36	90.0	29	72.5	34	85.0	4.502	0.105
CE2（36）	32	88.9	25	69.4	31	86.1	5.278	0.071
CE3（34）	29	85.3	22	64.7	28	82.4	4.828	0.089
CE4（43）	38	88.4	30	69.8	36	83.7	5.160	0.076
CE5（35）	31	88.6	24	68.6	29	82.9	4.643	0.098
合计 Total（188）	166	88.3	130	69.1^b	158	84.0	24.213	< 0.001
F组 F group
CE1（177）	161	91.0	133	75.1^b	154	87.0	18.193	< 0.001
CE2（173）	158	91.3	130	75.1^b	150	86.7	18.257	< 0.001
CE3（147）	127	86.4	108	73.5^b	124	84.4	9.378	0.009
CE4（265）	241	90.9	206	77.7^b	229	86.4	18.757	< 0.001
CE5（161）	145	90.1	116	72.0^b	138	85.7	19.801	< 0.001
合计 Total（923）	832	90.1	693	75.1^b	795	86.1^a	82.654	< 0.001
G组 G group
CE1（353）	318	91.0	288	81.6^b	321	90.9	17.292	< 0.001
CE2（354）	325	91.8	300	84.7^b	326	92.1	13.099	0.001
CE3（228）	196	86.0	173	75.9^b	201	88.2	14.084	< 0.001
CE4（415）	373	89.9	336	81.0^b	371	89.4	18.151	< 0.001
CE5（246）	220	89.4	190	77.2^b	220	89.4	19.524	< 0.001
合计 Total（1 596）	1 432	89.7	1 287	80.6^b	1 439	90.2	80.749	< 0.001

Table 4

Table 5

Table 6

References 22

[1]	Wen H, Vuitton L, Tuxun T, et al. Echinococcosis: Advances in the 21st century[J]. Clin Microbiol Rev, 2019, 32(2): e00075-18. [PubMed]
[2]	阿卜杜萨拉木•艾尼, 邵英梅, 吐尔干艾力•阿吉, 等. 中国棘球蚴病防诊治体系建设与创新实践[J]. 中国寄生虫学与寄生虫病杂志, 2019, 37(4): 388-394.
	Abudusalamu A, Shao YM, Tuerganaili A, et al. Establishment and innovative practice of Integrative System of Prevention, Diagnosis and Management for echinococcosis in China[J]. Chin J Parasitol Parasit Dis, 2019, 37(4): 388-394. (in Chinese)
[3]	李宏军. 肝包虫病影像学诊断专家共识[J]. 中华放射学杂志, 2021, 55(1): 5-11.
	Li HJ. Expert consensus on imaging diagnosis of hepatic echinococcosis[J]. Chin J Radiol, 2021, 55(1): 5-11. (in Chinese)
[4]	WHO Informal Working Group. International classification of ultrasound images in cystic echinococcosis for application in clinical and field epidemiological settings[J]. Acta Trop, 2003, 85(2): 253-261.
[5]	Ferrer Inaebnit E, Molina Romero FX, Segura Sampedro JJ, et al. A review of the diagnosis and management of liver hydatid cyst[J]. Rev Esp Enferm Dig, 2022, 114(1): 35-41.
[6]	侯娇, 温浩, 王明坤, 等. 肝细粒棘球蚴病手术患者病灶活性状态的影响因素分析[J]. 中国寄生虫学与寄生虫病杂志, 2022, 40(3): 309-315.
	Hou J, Wen H, Wang MK, et al. Analysis of the influencing factors of lesion activity in hepatic cystic echinococcosis patients[J]. Chin J Parasitol Parasit Dis, 2022, 40(3): 309-315. (in Chinese)
[7]	施亮, 熊春蓉, 刘毛毛, 等. 基于深度学习技术的湖北钉螺视觉智能识别模型效能评价[J]. 中国寄生虫学与寄生虫病杂志, 2021, 39(6): 764-770.
	Shi L, Xiong CR, Liu MM, et al. Evaluation of efficacy of visual intelligent recognition model for Oncomelania hupensis based on deep learning technology[J]. Chin J Parasitol Parasit Dis, 2021, 39(6): 764-770. (in Chinese)
[8]	Iqbal S, Qureshi AN, Li JQ, et al. On the analyses of medical images using traditional machine learning techniques and convolutional neural networks[J]. Arch Comput Meth Eng, 2023, 30(5): 3173-3233.
[9]	Wu M, Yan CB, Wang XR, et al. Automatic classification of hepatic cystic echinococcosis using ultrasound images and deep learning[J]. J Ultrasound Med, 2022, 41(1): 163-174.
[10]	王正业, 热娜古丽•艾合麦提尼亚孜, 王晓荣, 等. 肝囊型包虫病超声图影像区域分割算法研究[J]. 中国医疗设备, 2022, 37(10): 18-23, 28.
	Wang ZY, Renaguli AHMTNYZ, Wang XR, et al. Research on region segmentation algorithm of hepatic cystic echinococcosis ultrasonoscopy[J]. China Med Devices, 2022, 37(10): 18-23, 28. (in Chinese)
[11]	Ma T, Wang Q, Hao M, et al. Epidemiological characteristics and risk factors for cystic and alveolar echinococcosis in China: An analysis of a national population-base field survey[J]. Parasit Vectors. 2023, 16(1): 181.
[12]	韩帅, 李石柱. 我国棘球蚴病流行现状及重点防治工作思考[J]. 中国寄生虫学与寄生虫病杂志, 2025, 43(1): 1-5.
	Han S, Li SZ. Echinococcosis in China: Current status and future disease control priorities[J]. Chin J Parasitol Parasit Dis, 2025, 43(1): 1-5. (in Chinese)
[13]	Zi JR, Xiao D, Peng J, et al. Epidemiological survey of cystic echinococcosis in southwest China: From the Qinghai-Tibet plateau to the area of Yunnan[J]. World J Hepatol, 2024, 16(2): 229-240.
[14]	Suganyadevi S, Seethalakshmi V, Balasamy K. A review on deep learning in medical image analysis[J]. Int J Multimed Inf Retr, 2022, 11(1): 19-38.
[15]	Litjens G, Kooi T, Bejnordi BE, et al. A survey on deep learning in medical image analysis[J]. Med Image Anal, 2017, 42: 60-88.
[16]	Cao LL, Peng M, Xie X, et al. Artificial intelligence in liver ultrasound[J]. World J Gastroenterol, 2022, 28(27): 3398-3409.
[17]	Dahlblom V, Dustler M, Tingberg A, et al. Breast cancer screening with digital breast tomosynthesis: Comparison of different reading strategies implementing artificial intelligence[J]. Eur Radiol, 2023, 33(5): 3754-3765.
[18]	Vetter M, Waldner MJ, Zundler S, et al. Artificial intelligence for the classification of focal liver lesions in ultrasound: A systematic review[J]. Ultraschall Med, 2023, 44(4): 395-407.
[19]	Yousef R, Gupta G, Yousef N, et al. A holistic overview of deep learning approach in medical imaging[J]. Multimed Syst, 2022, 28(3): 881-914.
[20]	Nalla V, Pouriyeh S, Parizi RM, et al. Deep learning for computer-aided abnormalities classification in digital mammogram: A data-centric perspective[J]. Curr Probl Diagn Radiol, 2024, 53(3): 346-352.
[21]	Popa SL, Grad S, Chiarioni G, et al. Applications of artificial intelligence in the automatic diagnosis of focal liver lesions: A systematic review[J]. J Gastrointestin Liver Dis, 2023, 32(1): 77-85.
[22]	吴凯, 夏青, 贾立铭, 等. 人工智能辅助疟原虫检测平台的开发与应用效果研究[J]. 中国寄生虫学与寄生虫病杂志, 2025, 43(3): 345-350.
	Wu K, Xia Q, Jia LM, et al. Research on development and application of an artifical intelligence platform assisting Plasmodium detection[J]. Chin J Parasitol Parasit Dis, 2025, 43(3): 345-350. (in Chinese)

特征 Feature	病例数 No. case	构成比/% Proportion/%
年龄组/岁 Age group/years
< 40	291	30.6
40~59	440	46.2
≥ 60	221	23.2
性别 Gender
男 Male	509	53.5
女 Female	443	46.5
民族 Nationality
汉族 Han	390	41.0
维吾尔族 Weiwuer	266	27.9
哈萨克族 Hasake	188	19.7
回族 Hui	48	5.0
蒙古族 Menggu	31	3.3
藏族 Zang	19	2.0
锡伯族 Xibo	10	1.1
犬类接触史 History of exposure to dogs
有 Yes	425	44.6
无 No	527	55.4

评价指标 Evaluation metric	CE1	CE2	CE3	CE4	CE5
精确率/% Precision/%	0.946	0.898	0.838	0.908	0.870
召回率/% Recall/%	0.901	0.918	0.860	0.899	0.894
特异度/% Specificity/%	0.986	0.970	0.972	0.968	0.976
F1分数/% F1 score/%	0.923	0.908	0.849	0.904	0.882

组别 Group	Kappa
组别 Group	低年资医师与模型 Junior physicians and model	高年资医师与模型 Senior physicians and model	低年资与高年资医师 Junior and senior physicians
A	0.855	0.888	0.863
B	0.698	0.849	0.791
C	0.716	0.836	0.774
D	0.687	0.857	0.772
E	0.674	0.840	0.754
F	0.699	0.847	0.775
G	0.766	0.865	0.813