Usefulness of a deep learning system for diagnosing Sjögren’s syndrome using ultrasonography images

Abstract

Objectives: We evaluated the diagnostic performance of a deep learning system for the detection of Sjögren’s syndrome (SjS) in ultrasonography (US) images, and compared it with the performance of inexperienced radiologists. Methods: 100 patients with a confirmed diagnosis of SjS according to both the Japanese criteria and American-European Consensus Group criteria and 100 non-SjS patients that had a dry mouth and suspected SjS but were definitively diagnosed as non-SjS were enrolled in this study. All the patients underwent US scans of both the parotid glands (PG) and submandibular glands (SMG). The training group consisted of 80 SjS patients and 80 non-SjS patients, whereas the test group consisted of 20 SjS patients and 20 non-SjS patients for deep learning analysis. The performance of the deep learning system for diagnosing SjS from the US images was compared with the diagnoses made by three inexperienced radiologists. Results: The accuracy, sensitivity and specificity of the deep learning system for the PG were 89.5, 90.0 and 89.0%, respectively, and those for the inexperienced radiologists were 76.7, 67.0 and 86.3%, respectively. The deep learning system results for the SMG were 84.0, 81.0 and 87.0%, respectively, and those for the inexperienced radiologists were 72.0, 78.0 and 66.0%, respectively. The AUC for the inexperienced radiologists was significantly different from that of the deep learning system. Conclusions: The deep learning system had a high diagnostic ability for SjS. This suggests that deep learning could be used for diagnostic support when interpreting US images.

Machine Learning and Intelligent Diagnostics in Dental and Orofacial Pain Management: A Systematic Review

“…When assessing glandular disorders, radiologists demonstrated better agreement ( k = 0.65) for disorders of visibly larger glands (parotid) as opposed to smaller glands ( k = 0.51) obstructed by bony anatomy (submandibular gland) [ 13 ]. This deemed machine learning more sensitive to glandular anomalies but was also equally prone to making mistakes.…”

“…This deemed machine learning more sensitive to glandular anomalies but was also equally prone to making mistakes. Kise developed deep-learned systems to diagnose Sjogren's syndrome from both ultrasound imaging ( parotid gland : Ac = 0.89, Sn = 0.90, SP = 0.89; submandibular gland : Ac = 0.84, Sn = 0.81, Sp = 0.87) [ 13 ] and computed tomography (Ac = 0.96, Sn = 1.00, SP = 0.92) [ 12 ]. The authors found that only clinicians with >30 years of experience were able to compete (Ac = 0.98, Sn = 0.99, Sp = 0.97) with the deep learning algorithm (Ac = 0.96, Sn = 1.00, Sp = 0.92) in diagnosing salivary gland disorders from 3D CT images [ 12 ].…”

“…The authors found that only clinicians with >30 years of experience were able to compete (Ac = 0.98, Sn = 0.99, Sp = 0.97) with the deep learning algorithm (Ac = 0.96, Sn = 1.00, Sp = 0.92) in diagnosing salivary gland disorders from 3D CT images [ 12 ]. The outcomes for clinicians were, however, substantially poorer when made to diagnose 2D radiographs ( parotid gland : Ac = 0.77, Sn = 0.67, Sp = 0.86; submandibular gland : Ac = 0.72, Sn = 0.78, Sp = 0.66) [ 13 ]. Regardless, deep learning was shown to be a valuable diagnostic support for inexperienced clinicians (Ac = 0.77–0.84, Sn = 0.78, Sp = 0.75–0.89) [ 12 , 14 ] to accurately diagnose gland-related orofacial pain.…”

“…Maxillofacial fractures [ 8 ], as well as iatrogenic trauma/infection during dental restorative/endodontic treatment [ 2 ], may elicit varying levels of pain. Bone diseases [ 9 ], temporomandibular joint disorders [ 10 ], space infections [ 11 ], salivary gland disorders [ 12 , 13 ], and sinusitis [ 14 ] elicit pain. Furthermore, neuralgia and secondary sensory nerve compression due to growing cysts and tumors can elicit severe pain [ 15 , 16 ].…”

Application of Artificial Intelligence in Dentistry

“…Focusing on the expression of inflammatory cytokines genes, ANN, SVM, and RF were all capable of distinguishing oral lichen planus from other white lesions of the oral mucosa (Jeon et al 2015). When used for recognizing steatosis (i.e., abnormal retention of lipids) of the salivary gland parenchyma in ultrasonography images, ANN was superior to inexperienced radiologists in differentiating patients with true Sjögren’s syndrome from those with xerostomia (Kise et al 2020).…”