RESUMO
INTRODUCTION: Oral tumors necessitate a dependable computer-assisted pathological diagnosis system considering their rarity and diversity. A content-based image retrieval (CBIR) system using deep neural networks has been successfully devised for digital pathology. No CBIR system for oral pathology has been investigated because of the lack of an extensive image database and feature extractors tailored to oral pathology. MATERIALS AND METHODS: This study uses a large CBIR database constructed from 30 categories of oral tumors to compare deep learning methods as feature extractors. RESULTS: The highest average area under the receiver operating characteristic curve (AUC) was achieved by models trained on database images using self-supervised learning (SSL) methods (0.900 with SimCLR and 0.897 with TiCo). The generalizability of the models was validated using query images from the same cases taken with smartphones. When smartphone images were tested as queries, both models yielded the highest mean AUC (0.871 with SimCLR and 0.857 with TiCo). We ensured the retrieved image result would be easily observed by evaluating the top 10 mean accuracies and checking for an exact diagnostic category and its differential diagnostic categories. CONCLUSION: Training deep learning models with SSL methods using image data specific to the target site is beneficial for CBIR tasks in oral tumor histology to obtain histologically meaningful results and high performance. This result provides insight into the effective development of a CBIR system to help improve the accuracy and speed of histopathology diagnosis and advance oral tumor research in the future.
RESUMO
The clinical differences between odontogenic myxoma (OM) and odontogenic myxofibroma (OMF), and the clinical significance of their classifications, remain unclear. This study reviewed the clinicopathological characteristics of patients with OM or OMF and evaluated the fibrous component of the specimens. Medical records of 21 patients with OM or OMF who underwent tumour resection were reviewed. The percentage of fibrous tissue on the representative sections was evaluated using haematoxylin and eosin- and Masson's trichrome-stained specimens. Histopathological diagnoses included 11 OMs and 10 OMFs with no tumour recurrence except for two cases in which the dredging method was applied. More cortical bone perforation was observed in OM than in OMF cases, without significant differences. Location-locularity and apparent diffusion coefficient value (ADC)-cortical bone perforation were significantly correlated in all OM and OMF cases. The percentage of fibrous tissue in specimens showed bimodal distribution bordered by 45%. There was a significant association between diagnosis based on 45% fibrous tissue criterion and the final pathological diagnosis. Our study showed a tendency for cortical bone perforation in OM compared to OMF and correlation between ADC and cortical bone perforation. According to the histopathological analyses, the fibrous component of each case was bimodal with 45%, which may be a criterion to distinguish between OM and OMF. Accumulating knowledge, such as significant differences in prognosis, may allow for minimal surgical treatment options based on the diagnosis according to this novel histopathological criterion.
Assuntos
Fibroma , Mixoma , Tumores Odontogênicos , Humanos , Tumores Odontogênicos/patologia , Tumores Odontogênicos/cirurgia , Feminino , Masculino , Estudos Retrospectivos , Adulto , Pessoa de Meia-Idade , Mixoma/patologia , Mixoma/cirurgia , Fibroma/patologia , Fibroma/cirurgia , Idoso , Adolescente , Adulto Jovem , Diagnóstico DiferencialRESUMO
Objective: The accurate assessment of the involvement of mandibular gingival squamous cell carcinoma (SCC) is essential for determining the extent of resection and is also useful for predicting lymph node metastasis and prognosis. The purpose of this study was to investigate the factors for predicting the prognosis. Study design: We reviewed 134 patients with mandibular gingival SCC treated between 2008 and 2017. The clinical findings, TN stage, relationship between radiographical type and histological pattern, and factors affecting the survival rate were investigated. Results: The moth-eaten radiographic type was significantly associated with histologically infiltrative pattern. For all 134 cases, the 5-year OS was 89.5 %, and 5-year DSS was 93.9 %. The 5-year DSS was 95.0 % for cN0 and/or pN0 cases and 90.3 % for pN (+) cases, with a significant difference. The significant risk factors for lymph node metastasis were teeth extractions by previous physicians and moth-eaten radiographic type. Conclusion: The risk factor for poor prognosis was lymph node metastasis. In addition, teeth extractions by previous physicians and moth-eaten radiographic type were the risk factors for lymph node metastasis. It is recommended that these cases be treated considering the possibility of cervical lymph node metastasis.