Background Forests are an essential natural resource to humankind, providing a myriad of direct and indirect benefits. Natural disasters like forest fires have a major impact on global warming and the continued existence of life on Earth. Automatic identification of forest fires is thus an important field to research in order to minimize disasters. Early fire detection can also help decision-makers plan mitigation methods and extinguishing tactics. This research looks at fire/smoke detection from images using AI-based computer vision techniques. Convolutional Neural Networks (CNN) are a type of Artificial Intelligence (AI) approach that have been shown to outperform state-of-the-art methods in image classification and other computer vision tasks, but their training time can be prohibitive. Further, a pretrained CNN may underperform when there is no sufficient dataset available. To address this issue, transfer learning is exercised on pre-trained models. However, the models may lose their classification abilities on the original datasets when transfer learning is applied. To solve this problem, we use learning without forgetting (LwF), which trains the network with a new task but keeps the network’s preexisting abilities intact. Results In this study, we implement transfer learning on pre-trained models such as VGG16, InceptionV3, and Xception, which allow us to work with a smaller dataset and lessen the computational complexity without degrading accuracy. Of all the models, Xception excelled with 98.72% accuracy. We tested the performance of the proposed models with and without LwF. Without LwF, among all the proposed models, Xception gave an accuracy of 79.23% on a new task (BowFire dataset). While using LwF, Xception gave an accuracy of 91.41% for the BowFire dataset and 96.89% for the original dataset. We find that fine-tuning the new task with LwF performed comparatively well on the original dataset. Conclusion Based on the experimental findings, it is found that the proposed models outperform the current state-of-the-art methods. We also show that LwF can successfully categorize novel and unseen datasets.

Thyroid incidentalomas are common findings during imaging studies including 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) for cancer evaluation. Although the overall incidence of incidental thyroid uptake detected on PET imaging is low, clinical attention should be warranted owing to the high incidence of harboring primary thyroid malignancy. We retrospectively reviewed 2,368 dual-time-point 18F-FDG PET/CT cases that were undertaken for cancer evaluation from November 2007 to February 2009, to determine the clinical impact of dual-time-point imaging in the differential diagnosis of thyroid incidentalomas. Focal thyroid uptake was identified in 64 PET cases and final diagnosis was clarified with cytology/histology in a total of 27 patients with 18F-FDG-avid incidental thyroid lesion. The maximum standardized uptake value (SUVmax) of the initial image (SUV1) and SUVmax of the delayed image (SUV2) were determined, and the retention index (RI) was calculated by dividing the difference between SUV2 and SUV1 by SUV1 (i.e., RI = [SUV2 - SUV1]/SUV1 × 100). These indices were compared between patient groups that were proven to have pathologically benign or malignant thyroid lesions. There was no statistically significant difference in SUV1 between benign and malignant lesions. SUV2 and RI of the malignant lesions were significantly higher than the benign lesions. The areas under the ROC curves showed that SUV2 and RI have the ability to discriminate between benign and malignant thyroid lesions. The predictability of dual-time-point PET parameters for thyroid malignancy was assessed by ROC curve analyses. When SUV2 of 3.9 was used as cut-off threshold, malignancy on the pathology could be predicted with a sensitivity of 87.5 % and specificity of 75 %. A thyroid lesion that shows RI greater than 12.5 % could be expected to be malignant (sensitivity 88.9 %, specificity 66.3 %). All malignant lesions showed an increase in SUVmax on the delayed images compared with the initial images. But in the group of benign lesions, 37.5 % (6/16) showed a decrease or no change in SUVmax. Dual-time-point 18F-FDG PET/CT, obtaining additional images 2 h after injection, seems to be a complementary method for the differentiation between malignancy and benignity of incidental thyroid lesions.

The SUV-R performed well in distinguishing between metastatic and benign lymph nodes. In particular, SUV-R was found to have a better diagnostic performance than SUV-LN in the low SUV-T group.

Even though computer vision has been developing, edge detection is still one of the challenges in that field. It comes from the limitations of the complementary metal oxide semiconductor (CMOS) Image sensor used to collect the image data, and then image signal processor (ISP) is additionally required to understand the information received from each pixel and performs certain processing operations for edge detection. Even with/without ISP, as an output of hardware (camera, ISP), the original image is too raw to proceed edge detection image, because it can include extreme brightness and contrast, which is the key factor of image for edge detection. To reduce the onerousness, we propose a pre-processing method to obtain optimized brightness and contrast for improved edge detection. In the pre-processing, we extract meaningful features from image information and perform machine learning such as k-nearest neighbor (KNN), multilayer perceptron (MLP) and support vector machine (SVM) to obtain enhanced model by adjusting brightness and contrast. The comparison results of F1 score on edgy detection image of non-treated, pre-processed and pre-processed with machine learned are shown. The pre-processed with machine learned F1 result shows an average of 0.822, which is 2.7 times better results than the non-treated one. Eventually, the proposed pre-processing and machine learning method is proved as the essential method of pre-processing image from ISP in order to gain better edge detection image. In addition, if we go through the pre-processing method that we proposed, it is possible to more clearly and easily determine the object required when performing auto white balance (AWB) or auto exposure (AE) in the ISP. It helps to perform faster and more efficiently through the proactive ISP.

Purpose The aim of this study was to evaluate the relationship between semiquantitative parameters on 18 F-FDG PET/CT including maximum standardized uptake value (SUVmax), mean standardized uptake value (SUVmean), metabolic tumor volume (MTV), and total lesion glycolysis (TLG) and the expression level of Ki-67 in small-cell lung cancer (SCLC). Methods Ninety-four consecutive patients with SCLC were enrolled in this study. They underwent 18 F-FDG PET/CT for initial evaluation of SCLC, and we measured SUVmax, avgSUVmean, MTVsum, and TLGtotal on 18 F-FDG PET/ CT images. The protein expression of Ki-67 was examined by immunohistochemical staining. Results Significant correlations were found between the MTVsum and Ki-67 labeling index (r=0.254, p=0.014) and the TLGtotal and Ki-67 labeling index (r=0.239, p=0.020). No correlation was found between the SUVmax and Ki-67 labeling index (r=0.116, p=0.264) and the avgSUVmean and Ki-67 labeling index (r=0.031, p=0.770). Dividing the Ki-67 expression level into three categories, it was suggested that increasing Ki-67 expression level caused a stepwise increase in the MTVsum and TLGtotal. (p=0.028 and 0.039, respectively), but not the SUVmax and avgSUVmean (p= 0.526 and 0.729, respectively). Conclusion In conclusion, the volume-based parameters of 18 F-FDG PET/CT correlate with immunohistochemical staining of Ki-67 in SCLC. Measurement of the MTVsum and TLGtotal by 18 F-FDG PET/CT might be a simple, noninvasive, and useful method to determine the proliferative potential of cancer cells.