RESUMO
Automation of medical data analysis is an important topic in modern cancer diagnostics, aiming at robust and reproducible workflows. Therefore, we used a dataset of breast US images (252 malignant and 253 benign cases) to realize and compare different strategies for CAD support in lesion detection and classification. Eight different datasets (including pre-processed and spatially augmented images) were prepared, and machine learning algorithms (i.e., Viola-Jones; YOLOv3) were trained for lesion detection. The radiomics signature (RS) was derived from detection boxes and compared with RS derived from manually obtained segments. Finally, the classification model was established and evaluated concerning accuracy, sensitivity, specificity, and area under the Receiver Operating Characteristic curve. After training on a dataset including logarithmic derivatives of US images, we found that YOLOv3 obtains better results in breast lesion detection (IoU: 0.544 ± 0.081; LE: 0.171 ± 0.009) than the Viola-Jones framework (IoU: 0.399 ± 0.054; LE: 0.096 ± 0.016). Interestingly, our findings show that the classification model trained with RS derived from detection boxes and the model based on the RS derived from a gold standard manual segmentation are comparable (p-value = 0.071). Thus, deriving radiomics signatures from the detection box is a promising technique for building a breast lesion classification model, and may reduce the need for the lesion segmentation step in the future design of CAD systems.
RESUMO
BACKGROUND: Because of bisphosphonate medication, dental implantation with a subsequent infection poses a relevant risk factor to suffer from medication-related osteonecrosis of the jaw. This rat study evaluated different implant materials under systemic bisphosphonate delivery using micro-computed tomography (µCT) images. METHODS: Fifty-four rats were randomly allocated into a control group 1, test group 2 with intravenous drug application of zoledronic acid and test group 3 with a subcutaneous application of alendronic acid. After 4 weeks of drug delivery, the first molar on each side of the upper jaw was extracted, and either a zirconia or a titanium implant was immediately inserted. Radiological examinations at four timepoints before the operation, 1 week later, 6 weeks later and after 12 weeks of follow up included µCT measurements of the in vivo peri-implant bone loss. µCT measurements of the ex vivo peri-implant bony structure after 12 weeks follow-up covered the bone mineral density, -volume, -trabecular thickness and -separation. RESULTS: Both test groups showed a significant increase in bone loss over time (P < 0.05). The clinical observations of exposed bone revealed that most cases occurred under alendronic acid delivery. Exposed bone was recorded only in the test groups around both titanium and zirconia implants. Regarding the peri-implant bony structure, no significant differences were found between both materials. CONCLUSIONS: Systemic bisphosphonate delivery led to increased peri-implant bone loss over time after immediate implant insertion. In terms of bone resorption and bone quality parameters, no implant material was superior to the other.