Comparison of Explainable Artificial Intelligence Model and Radiologist Review Performances to Detect Breast Cancer in 752 Patients.

OBJECTIVES: Breast cancer is a type of cancer caused by the uncontrolled growth of cells in the breast tissue. In a few cases, erroneous diagnosis of breast cancer by specialists and unnecessary biopsies can lead to various negative consequences. In some cases, radiologic examinations or clinical findings may raise the suspicion of breast cancer, but subsequent detailed evaluations may not confirm cancer. In addition to causing unnecessary anxiety and stress to patients, such diagnosis can also lead to unnecessary biopsy procedures, which are painful, expensive, and prone to misdiagnosis. Therefore, there is a need for the development of more accurate and reliable methods for breast cancer diagnosis. METHODS: In this study, we proposed an artificial intelligence (AI)-based method for automatically classifying breast solid mass lesions as benign vs malignant. In this study, a new breast cancer dataset (Breast-XD) was created with 791 solid mass lesions belonging to 752 different patients aged 18 to 85 years, which were examined by experienced radiologists between 2017 and 2022. RESULTS: Six classifiers, support vector machine (SVM), K-nearest neighbor (K-NN), random forest (RF), decision tree (DT), logistic regression (LR), and XGBoost, were trained on the training samples of the Breast-XD dataset. Then, each classifier made predictions on 159 test data that it had not seen before. The highest classification result was obtained using the explainable XGBoost model (X2GAI) with an accuracy of 94.34%. An explainable structure is also implemented to build the reliability of the developed model. CONCLUSIONS: The results obtained by radiologists and the X2GAI model were compared according to the diagnosis obtained from the biopsy. It was observed that our developed model performed well in cases where experienced radiologists gave false positive results.

Long term results of glass ionomer ossiculoplasty.

OBJECTIVE: To analyze and compare the early and late post-operative results of glass ionomer bone cement (GIBC) used in ossiculoplasty. METHODS: The pre-operative, early post-operative (3 months) and late post-operative (> 2 years) audiometric findings, namely the pure-tone average (PTA), bone conduction threshold (BCT) and air-bone gap (ABG) of 40 patients who underwent GIBC ossiculoplasty for different etiologies were analyzed. Early and late results were compared. Also, the patients were grouped in terms of prognostic factors and applied ossiculoplasty techniques, and the results were compared between the groups. RESULTS: There were statistically significant improvements in the pure-tone average and air-bone gap of the patients in the early post-operative period (PTA from 59.60 ± 15.95 to 40.37 ± 17.83 and ABG from 37.12 ± 11.18 to 19.78 ± 10.41, p < 0.001 for both). There were no statistically significant changes in any of the audiometric parameters in the late post-operative period (PTA from 40.37 ± 17.83 to 39.79 ± 17.91, ABG from 19.78 ± 10.41 to 19.32 ± 9.60, BCT from 17.99 ± 12.71 to 18.31 ± 13.99, p > 0.05 for all). Presence of tympanosclerosis was found to be the only prognostic factor to affect the outcome. CONCLUSION: GIBC is a safe and reliable material for ossiculoplasty, which maintains its ability to conduct sound in the long-term follow-up.

An Explainable Vision Transformer Model Based White Blood Cells Classification and Localization.

White blood cells (WBCs) are crucial components of the immune system that play a vital role in defending the body against infections and diseases. The identification of WBCs subtypes is useful in the detection of various diseases, such as infections, leukemia, and other hematological malignancies. The manual screening of blood films is time-consuming and subjective, leading to inconsistencies and errors. Convolutional neural networks (CNN)-based models can automate such classification processes, but are incapable of capturing long-range dependencies and global context. This paper proposes an explainable Vision Transformer (ViT) model for automatic WBCs detection from blood films. The proposed model uses a self-attention mechanism to extract features from input images. Our proposed model was trained and validated on a public dataset of 16,633 samples containing five different types of WBCs. As a result of experiments on the classification of five different types of WBCs, our model achieved an accuracy of 99.40%. Moreover, the model's examination of misclassified test samples revealed a correlation between incorrect predictions and the presence or absence of granules in the cell samples. To validate this observation, we divided the dataset into two classes, Granulocytes and Agranulocytes, and conducted a secondary training process. The resulting ViT model, trained for binary classification, achieved impressive performance metrics during the test phase, including an accuracy of 99.70%, recall of 99.54%, precision of 99.32%, and F-1 score of 99.43%. To ensure the reliability of the ViT model's, we employed the Score-CAM algorithm to visualize the pixel areas on which the model focuses during its predictions. Our proposed method is suitable for clinical use due to its explainable structure as well as its superior performance compared to similar studies in the literature. The classification and localization of WBCs with this model can facilitate the detection and reporting process for the pathologist.

An Explainable Deep Learning Model to Prediction Dental Caries Using Panoramic Radiograph Images.

Dental caries is the most frequent dental health issue in the general population. Dental caries can result in extreme pain or infections, lowering people's quality of life. Applying machine learning models to automatically identify dental caries can lead to earlier treatment. However, physicians frequently find the model results unsatisfactory due to a lack of explainability. Our study attempts to address this issue with an explainable deep learning model for detecting dental caries. We tested three prominent pre-trained models, EfficientNet-B0, DenseNet-121, and ResNet-50, to determine which is best for the caries detection task. These models take panoramic images as the input, producing a caries-non-caries classification result and a heat map, which visualizes areas of interest on the tooth. The model performance was evaluated using whole panoramic images of 562 subjects. All three models produced remarkably similar results. However, the ResNet-50 model exhibited a slightly better performance when compared to EfficientNet-B0 and DenseNet-121. This model obtained an accuracy of 92.00%, a sensitivity of 87.33%, and an F1-score of 91.61%. Visual inspection showed us that the heat maps were also located in the areas with caries. The proposed explainable deep learning model diagnosed dental caries with high accuracy and reliability. The heat maps help to explain the classification results by indicating a region of suspected caries on the teeth. Dentists could use these heat maps to validate the classification results and reduce misclassification.

The utility of acoustic pharyngometry in treatment of obstructive sleep apnea patients with expansion sphincter pharyngoplasty surgery.

PURPOSE: Expansion sphincter pharyngoplasty (ESP) is a common surgery for patients with obstructive sleep apnea (OSA) which aims to correct the obstruction at the palatal level. The effectiveness of ESP has been widely shown in the literature using surgical success rates, but to our knowledge, there is no research which documents the changes in the upper airway anatomy objectively. We aimed to demonstrate the effectiveness of expansion sphincter pharyngoplasty using acoustic pharyngometry. We also aimed to study the possible utility of acoustic pharyngometry in predicting surgical outcomes. METHODS: Pre- and post-operative acoustic pharyngometry and polysomnography data of patients who underwent expansion sphincter pharyngoplasty were compared prospectively. Minimum cross-sectional area (MCA) and total volume of the pharynx (TPV), apnea-hypopnea index (AHI), and surgical success rates were evaluated. RESULTS: Fifty-two patients with OSA were invited to this study, and 35 patients who agreed to participate were enrolled. All patients underwent ESP surgery. Surgical success rate was 63% according to Sher's criteria. The mean AHI of the patients decreased from 29.6 ± 16.3 to 18.3 ± 18.1. MCA increased from 1.1 ± 0.4 to 2.3 ± 0.4 cm2, and TPV increased from 21.1 ± 6.9 to 31.7 ± 5.5 cm3. Comparative analysis of the successful and unsuccessful groups yielded no significant differences between the groups concerning pre- and post-operative MCA and TPV or in mean changes in MCA and TPV achieved with the surgery. CONCLUSION: Improvement in the upper airway anatomy by expansion sphincter pharyngoplasty can be clearly demonstrated using acoustic pharyngometry. Acoustic pharyngometry findings are quite similar in patients with successful and unsuccessful outcomes; therefore, pharyngometry findings cannot be used to predict surgical success; and surgical success cannot be solely attributed to the changes in MCA and TPV.

Supine or non-supine apnea; which can be treated better with expansion sphincter pharyngoplasty?

OBJECTIVES: Expansion sphincter pharyngoplasty is a well-known palatal correction surgery applied for obstructive sleep apnea patients. Even though high success rates are reported, in some patients, the desired outcome cannot be achieved. Therefore, patient selection is crucial to avoid undesirable outcomes. In this study we present our results with expansion sphincter pharyngoplasty, and we aim to show the difference between supine and non-supine apnea in terms of their response to the surgery, in order to utilize position dependency for selection of surgical candidates. METHODS: Pre- and post-operative polysomnography results of patients who underwent expansion sphincter pharyngoplasty were analyzed retrospectively. Total AHI, supine AHI, non-supine AHI, supine/non-supine AHI ratio and surgical success values are compared. RESULTS: 85 patients were included to the study. Mean AHI significantly decreased from 48.7 ± 27.99 to 26.37 ± 21.16 with the surgery. Surgical success rate was found to be 51.8%. Both supine and non-supine apnea decreased significantly with the surgery, but the decrease was significantly higher in non-supine apnea (20.6% to 39.1% respectively, p = 0.016). There was significant negative correlation between pre-operative supine to non-supine AHI ratio and the change in AHI, showing that supine dominant patients had less improvement with the surgery (r = 0.274, p = 0.01). CONCLUSION: Expansion sphincter pharyngoplasty is an effective surgery which achieves significant improvement in AHI. Non-supine respiratory events respond better to the surgery than supine events.

Role of indocyanine green combined with radiotracer-Technetium 99 m in neck surgery for primary and recurrent head and neck cancer: preliminary results of a tertiary cancer center.

PURPOSE: Neck dissection (ND), whether therapeutic or elective, is an essential component of the treatment of head and neck squamous cell carcinoma (HNSCC). Due to altered anatomy and fibrosis caused by initial treatments, surgeons face challenges during salvage ND. A combination of Technetium-99 m and indocyanine green (ICG) previously used in the sentinel lymph node (SLN) biopsy for oral cavity cancers, may be useful in different types of neck surgeries. We aimed to show the additional advantage of this combination in detecting HNSCC and thyroid cancer recurrences, as well as individual lymphatic drainage in elective ND. METHODS: We retrospectively reviewed medical records of patients, underwent neck surgery guided with ICG and Tc-99 m, in Hacettepe University Hospital between June 2018 and December 2020. In a total of 28 patients, intraoperative gamma probes were paired with near infrared (NIR) cameras. Patients are divided into two groups: neck surgery with recurrent occult lesion localization (NS- ROLL) (n: 14) and ND with SLN screening (ND-SLNS) (n: 14). RESULTS: Among all 14 patients in NS-ROLL group, recurrent diseases, hidden behind tissues were 100% successfully localized. In ND-SLNS group, 238 lymph nodes were harvested, metastasis rate was 31.3% (10/32) in sentinel nodes. SLNS revealed 100% accuracy in detecting metastasis in clinically N0 neck (10/238). Contralateral lymphatic drainage was observed in three patients (lateral-sided oral cavity SCC). In two patients (floor of mouth), three sentinel nodes were detected by NIR only. CONCLUSION: The use of ICG-radiotracer provides additional value in disease removal for both primary and recurrent tumors of the head and neck.

Unilateral Facial Paralysis and Ophthalmoplegia Caused by Lower Lip Carcinoma: A Case Report.

We discuss a case of lower lip carcinoma which presented with atypical symptoms; facial paralysis, conductive type hearing loss, and ophthalmoplegia. Due to an earlier resection, no mass was evident on the primary examination. Diagnostic imaging revealed a mass originating from the lower lip, the perineural spread of the tumor along the left inferior alveolar nerve to the left infratemporal fossa and the left foramen ovale. Through a retrograde course from the foramen ovale, the tumor extended the ipsilateral cavernous sinus, Meckel's cave, and cisternal portion of the CN V. This atypical spread pattern of the tumor caused symptoms that may be attributed to a diagnosis related to the ear. The biopsy confirmed squamous cell carcinoma, and the patient was referred for chemotherapy and radiotherapy.

Results of coblation midline glossectomy for obstructive sleep apnea.

OBJECTIVE: The aim of this study is to determine the effectiveness of coblation midline glossectomy for obstructive sleep apnea (OSA) when used as an isolated procedure. We also aim to compare the effect of this surgical procedure on supine and non-supine apnea. MATERIALS AND METHODS: The medical records of patients who underwent isolated tongue base surgery as a part of step-wise surgeries between January 2014 and February 2019 are retrospectively reviewed. Pre-operative and post-operative Epworth sleepiness score (ESS), body mass index (BMI), and polysomnographic data, including the apnea-hypopnea index (AHI), oxygen desaturation index (ODI), supine AHI, and non-supine AHI of the patients were compared. RESULTS: The study included 29 patients (26 male and 3 female). AHI improved significantly, decreasing from 34.9 ± 20.9 to 25.8 ± 17.6. Supine AHI decreased from 62.55 ± 28.23 to 55.18 ± 31.67 post-operatively, but this decrease was not significant. Non-supine AHI decreased significantly from 22.49 ± 24.02 to 14.08 ± 17.46. ESS and ODI also improved significantly. CONCLUSION: Coblation midline glossectomy is an effective surgical procedure when applied solely, with a success rate of 52%. Non-supine apnea benefits to a greater degree than supine apnea from this surgical procedure.