All-star approach to a small medical imaging dataset: combined deep, transfer, and classical machine learning approaches for the determination of radial head fractures.

BACKGROUND: Radial head fractures are often evaluated in emergency departments and can easily be missed. Automated or semi-automated detection methods that help physicians may be valuable regarding the high miss rate. PURPOSE: To evaluate the accuracy of combined deep, transfer, and classical machine learning approaches on a small dataset for determination of radial head fractures. MATERIAL AND METHODS: A total of 48 patients with radial head fracture and 56 patients without fracture on elbow radiographs were retrospectively evaluated. The input images were obtained by cropping anteroposterior elbow radiographs around a center-point on the radial head. For fracture determination, an algorithm based on feature extraction using distinct prototypes of pretrained networks (VGG16, ResNet50, InceptionV3, MobileNetV2) representing four different approaches was developed. Reduction of feature space dimensions, feeding the most relevant features, and development of ensemble of classifiers were utilized. RESULTS: The algorithm with the best performance consisted of preprocessing the input, computation of global maximum and global mean outputs of four distinct pretrained networks, dimensionality reduction by applying univariate and ensemble feature selectors, and applying Support Vector Machines and Random Forest classifiers to the transformed and reduced dataset. A maximum accuracy of 90% with MobileNetV2 pretrained features was reached for fracture determination with a small sample size. CONCLUSION: Radial head fractures can be determined with a combined approach and limitations of the small sample size can be overcome by utilizing pretrained deep networks with classical machine learning methods.

Comparison of subpectoral versus dual-plane implant based immediate breast reconstruction after nipple-areola sparing mastectomy.

BACKGROUND: Selection of implant pocket and size is a dilemma for surgeons especially if radiation therapy is envisaged after implant based immediate breast reconstruction (IBR). The aim of this study is to compare complication rates between subpectoral and the dual plane polyglactin mesh supported IBR after nipple-areola sparing mastectomy. PATIENTS AND METHOD: Reconstructive analysis of 208 breasts of 190 patients with breast cancer undergoing implant based IBR at a single university hospital were evaluated. The patients were reconstructed with either dual-plane polyglactin mesh supported (n=91) or subpectoral (n=117) implant based IBR after nipple-areola sparing mastectomy. Demographic data, and postoperative complications were compared. RESULTS: The mean age was 43.3 years. Early complications encountered in 12% (n=25) and late complications occurred in 18% (n=37) of breasts. Both early (13,7% vs 9.9%, P=0.406) and late complications (24.8% vs 8.8%, P=0.003) were more common in subpectoral group, but only late complication occurrence rate was statistically significant. Capsular contracture (P=0.000), inframammary fold problems (P=0.010), bottoming-out (P=0.370), mechanical shift (P=0.036) and animation deformity (P=0.007) were all more common in subpectoral group. Only rippling deformity (P=0.011) was more common in dual plane group. CONCLUSIONS: Dual plane IBR has acceptable complication rates compared to subpectoral IBR. It is associated with less capsular contracture, fewer animation and bottoming-out deformity and better inframammary fold appearance.

The critical shoulder angle, the acromial index, the glenoid version angle and the acromial angulation are associated with rotator cuff tears.

PURPOSE: To compare the critical shoulder angle (CSA), acromion index (AI), acromion angulation (AA) and glenoid version angle (GVA) between patients with full-thickness rotator cuff tears (RCTs) and patients with intact rotator cuffs. METHODS: Between 2014 and 2018, the CSA, AI, AA and GVA were measured in consecutively included patients aged > 40 years who underwent shoulder arthroscopy for full-thickness RCTs. A total of 437 patients with RCTs and a mean age of 51.2 years (± 5.8) were included, 35.7% of whom were male. In the control group, there were n = 433 patients (36.3% male) with an intact rotator cuff, and the mean age was 50.7 years (± 5.3). RESULTS: The mean AI for the RCT group was 0.7 ± 0.1, which was significantly higher than the mean AI for the control group (0.6 ± 0.1, p < 0.001). The mean CSA for the RCT group was 33.6° ± 3.9°, which was significantly higher than the mean CSA for the control group (31.5° ± 4°, p < 0.001). The mean AA for the RCT group was 13.9° ± 9°, which was significantly higher than the mean AA for the control group (12.4 ± 8.6, p = 0.012). The mean GVA for the RCT group was - 3.5° ± 4.6° and significantly retroverted compared with the mean GVA for the control group (- 2.2° ± 4.6°, p < 0.001). The cutoff values determined by the ROC curve analyses were as follows: 0.6 for AI, 31.4° for CSA, 9.6° for AA and - 2.6° for GVA. CONCLUSION: The CSA, AI, GVA and AA values measured by MRI were determined to be significantly related to full-thickness rotator cuff ruptures. The AI, CSA, AA and GVA may be considered risk factors for degenerative rotator cuff tears. Assessing the CSA, AI, GVA and AA can be helpful for diagnostic evaluation of patients with full-thickness RCTs. LEVEL OF EVIDENCE: III.

Evaluation of oxaliplatin-induced pulmonary toxicity in rats.

INTRODUCTION: The mechanism of oxaliplatin (OXA) induced pulmonary toxicity is not fully understood. AIM OF THE STUDY: The present study was designed to investigate the pulmonary toxicity of OXA that has been reported in previous studies. Study design: animal experiments. MATERIAL AND METHODS: A total of 40 female Wistar rats were divided into 5 groups. In group 1, 5% glucose was injected intra-peritoneally; then the rats were sacrificed on day 14. OXA was administered in groups 2, 3, 4, and 5; then the animals were sacrificed on day 7 in group 2, day 14 in group 3, day 28 in group 4 and day 48 in group 5. The groups were further categorized as short-term administration and long-term administration groups. Furthermore, tissue glutathione peroxidase (GPX) activity was measured in all rats. RESULTS: The mean GPX activities were 0.66 U/mg in the sham group, 0.74 U/mg in the short-term groups, and 0.74 U/mg in the long-term groups. We found that long-term OXA administration causes pulmonary toxicity resulting in increased intra-alveolar/interstitial macrophages and interstitial pneumonia. Similarly, we found reduced and permanent tissue GPX activity in rats that received OXA in higher doses and for a long term. CONCLUSIONS: Long-term OXA therapy causes toxic changes in the lung tissue.

Preoperative chemotherapy for T2 breast cancer is associated with improved surgical outcome.

BACKGROUND: The aim of this study is to compare the clinical outcome in T2 breast cancer patients who underwent preoperative chemotherapy (PC) and who did not. The study also tried to define a subgroup of patients, who are more beneficial after PC in terms of lower re-excision rates, better cosmetic results and local recurrence free survival. MATERIALS AND METHODS: 251 consecutive patients treated for nonmetastatic T2 invasive breast cancer were analyzed retrospectively. Of those; 141 underwent primary surgery (PS) followed by chemotherapy, whereas 110 were treated with combination of PC and surgery. RESULTS: The patients who were treated with PC had a significantly higher incidence of negative margins and lower rate of re-excision (5% vs. 16%, p = 0.02). Of all patients attempted breast conserving surgery (BCS), patients in the PC group were more likely to undergo BCS as their definitive operation compared to patients with PS group (BCS rates; PC group: 99% vs. PS group: 92%, p = 0.05). Multifocal disease (OR: 7, 95% Cl, 2.7-18.4, p = 0.0001) and PC (OR = 0.2; 95% CI, 0.06-0.72, p = 0.01) were factors associated with margin positivity in patients treated with BCS. There was no statistically significant difference in 5 year local-recurrence free survival rates between 2 groups. CONCLUSIONS: Our study shows that PC significantly decreases the re-excision in patients undergoing BCS with primary T2 breast tumors. This data suggests that any patient with a tumor greater than 2 cm might be considered for PC to increase BCS success with final negative margins.

Effects of different irrigation solutions on root dentine microhardness, smear layer removal and erosion.

This study aimed to compare the effects of different irrigants on root dentine microhardness, erosion and smear layer removal. A total of 72 root dentine slices were divided into six groups, according to the final irrigants used: Group 1: 17% ethylenediamine tetra-acetic acid (EDTA) + 2.5% NaOCl, Group 2: 7% maleic acid (MA) + 2.5% sodium hypochloride (NaOCl), Group 3: 1.3% NaOCl + mixture of tetracycline, acid and detergent (MTAD), Group 4: Smear Clear + 2.5% NaOCl, Group 5: 5% NaOCl, Group 6: saline. Vickers microhardness values were measured before and after treatment. In total, 42 root-halves were prepared for scanning electron microscope to evaluate the amount of smear and erosion in the coronal, middle and apical thirds. Data were analysed using two-way anova, Duncan and two-proportion z-tests. Maleic acid showed the greatest reduction in dentine microhardness (P < 0.05), followed by EDTA and MTAD. EDTA, maleic acid, MTAD and Smear Clear removed smear layer efficiently in the coronal and middle thirds of root canal. However, in the apical region, maleic acid showed more efficient removal of the smear layer than the other irrigants (P < 0.05).