Development and validation of the diagnostic accuracy of artificial intelligence-assisted ultrasound in the classification of splenic trauma

Abstract

Background: Spleen is the most vulnerable organ in abdominal trauma. Ultrasound (US) has become an important examination method for splenic trauma. However, the sensitivity of routine US in the diagnosis of splenic trauma is low. Contrast-enhanced ultrasound (CEUS) can improve the sensitivity, but it also has some limitations. This study sought to explore the application value of artificial intelligence (AI)-assisted US in the classification of splenic trauma. Methods:The splenic injuries of Bama miniature pigs were established. A large number of ultrasonic images were collected. Then, 3-fold cross validation (CV) was used to establish the animal models. Next, clinical ultrasonic images were collected at multiple centers. All injuries were diagnosed by CEUS, enhanced CT or surgery. We used animal models to fine tune a small amount of human data, and then established the final AI splenic trauma recognition model. The whole model was constructed by averaging the prediction ability of the 3 fine-tuned models. Finally, 2 doctors' recognition US results of splenic trauma were compared to the AI recognition results. The area under the curve (AUC), sensitivity, specificity, negative predictive value, and positive predictive value were used to evaluate the diagnostic performance in diagnosis of spleen trauma.Results: (I) Based on the receiver operating characteristic (ROC) curves, the test cohort 1 (AUC =0.90) and 2 (AUC =0.84) had a similar performance. Based on the decision curve analysis (DCA) curves, while threshold smaller than 0.8, the proposed model had better performance on test cohort 1 than test cohort 2.Test cohort 1 had higher sensitivity (0.82 vs. 0.71, P<0.01) and higher specificity (0.88 vs. 0.81, P<0.01) than test cohort 2. (II) The diagnostic accuracy of the AI model was higher than that of doctor 1 (0.82 vs. 0.62, P<0.001) and doctor 2 (0.82 vs. 0.66, P<0.001), and its specificity was higher than that of doctor (0.88 vs. 0.78, P=0.001).Conclusions: AI-assisted US diagnosis of splenic trauma can significantly improve the ultrasonic diagnosis rate. We still need to increase the number of samples to further improve the diagnostic efficiency of the model.

Applications of Deep Learning Algorithms to Ultrasound Imaging Analysis in Preclinical Studies on In Vivo Animals

“…Following the review of the title and abstract and, upon necessity, the full text further 187 records were rejected. After the reviewing process, a total of 36 original papers [ 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 , 61 , 62 , 63 ] and 20 conference proceedings [ 64 , 65 , 66 , 67 , 68 , 69 , 70 , 71 , 72 , 73 , 74 , 75 , 76 , 77 , 78 , 79 , 80 , 81 , 82 ,…”

“…The US is widely adopted for investigating organs and soft tissues in the abdominal cavity. Indeed, 14 (25%) of the reviewed papers [ 29 , 37 , 44 , 47 , 49 , 50 , 53 , 60 , 62 , 67 , 69 , 72 , 76 , 81 ] have been focused on abdominal US imaging. In particular, studies focused on liver, kidney, spleen and bowel examinations.…”

“…Eight papers [ 29 , 37 , 47 , 53 , 60 , 67 , 76 , 81 ] proposed DL models for classification tasks in organ disease. In particular, Banzato et al [ 29 ] developed a deep neural network using transfer learning (AlexNet retrained and fine-tuned) for the diagnosis of degenerative liver disease in dogs.…”

“…Moreover, in four papers, the authors proposed CNN models to classify in the rabbit the severity of liver fat content [ 53 , 67 , 76 ] and fibrosis [ 60 ]. Finally, Jiang et al [ 47 ] applied MobileNetV2 for 2D feature extraction, followed by ResNet models for the classification of splenic trauma in pigs. In the papers [ 37 , 50 , 60 ], all the authors proposed DL systems to classify the stages of liver fibrosis.…”

“…Alternatively, more complicated techniques are also developing, such as Generative Adversarial Networks (GANs), which generate plausible new data compared to the available ones [ 88 ]. The majority of the selected papers [ 28 , 29 , 30 , 31 , 32 , 33 , 34 , 36 , 37 , 38 , 41 , 42 , 45 , 47 , 48 , 49 , 50 , 57 , 58 , 61 , 71 , 72 , 73 , 74 , 77 , 78 , 79 , 80 , 82 ] used TL techniques to overcome data limitations rather than developing models from scratch. Data augmentation was performed in twenty-eight of the analysed papers, dealing with the cardiovascular system [ 31 , 34 , 48 , 56 , 73 , 75 , 77 ], abdominal organs [ 29 , 44 , 49 , 50 , 60 , 72 , 81 ], musculoskeletal system [ 55 , 61 ], embryo [ 28 , 57 , 78 , 79 , 80 , 82 ], breast tumour vascularisation [ 32 ,…”