Deep Learning Networks for Breast Lesion Classification in Ultrasound Images: A Comparative Study.

Accurate lesion classification as benign or malignant in breast ultrasound (BUS) images is a critical task that requires experienced radiologists and has many challenges, such as poor image quality, artifacts, and high lesion variability. Thus, automatic lesion classification may aid professionals in breast cancer diagnosis. In this scope, computer-aided diagnosis systems have been proposed to assist in medical image interpretation, outperforming the intra and inter-observer variability. Recently, such systems using convolutional neural networks have demonstrated impressive results in medical image classification tasks. However, the lack of public benchmarks and a standardized evaluation method hampers the performance comparison of networks. This work is a benchmark for lesion classification in BUS images comparing six state-of-the-art networks: GoogLeNet, InceptionV3, ResNet, DenseNet, MobileNetV2, and EfficientNet. For each network, five input data variations that include segmentation information were tested to compare their impact on the final performance. The methods were trained on a multi-center BUS dataset (BUSI and UDIAT) and evaluated using the following metrics: precision, sensitivity, F1-score, accuracy, and area under the curve (AUC). Overall, the lesion with a thin border of background provides the best performance. For this input data, EfficientNet obtained the best results: an accuracy of 97.65% and an AUC of 96.30%.Clinical Relevance- This study showed the potential of deep neural networks to be used in clinical practice for breast lesion classification, also suggesting the best model choices.

Comparative Analysis of Current Deep Learning Networks for Breast Lesion Segmentation in Ultrasound Images.

Automatic lesion segmentation in breast ultrasound (BUS) images aids in the diagnosis of breast cancer, the most common type of cancer in women. Accurate lesion segmentation in ultrasound images is a challenging task due to speckle noise, artifacts, shadows, and lesion variability in size and shape. Recently, convolutional neural networks have demonstrated impressive results in medical image segmentation tasks. However, the lack of public benchmarks and a standardized evaluation method hampers the networks' performance comparison. This work presents a benchmark of seven state-of-the-art methods for the automatic breast lesion segmentation task. The methods were evaluated on a multi-center BUS dataset composed of three public datasets. Specifically, the U-Net, Dynamic U-Net, Semantic Segmentation Deep Residual Network with Variational Autoencoder (SegResNetVAE), U-Net Transformers, Residual Feedback Network, Multiscale Dual Attention-Based Network, and Global Guidance Network (GG-Net) architectures were evaluated. The training was performed with a combination of the cross-entropy and Dice loss functions and the overall performance of the networks was assessed using the Dice coefficient, Jaccard index, accuracy, recall, specificity, and precision. Despite all networks having obtained Dice scores superior to 75%, the GG-Net and SegResNetVAE architectures outperform the remaining methods, achieving 82.56% and 81.90%, respectively. Clinical Relevance- The results of this study allowed to prove the potential of deep neural networks to be used in clinical practice for breast lesion segmentation also suggesting the best model choices.

Novel laser-induced luminescence resulting from benzophenone/O-propylated p-tert-butylcalix[4]arene complexes. A diffuse reflectance study.

Laser-induced room temperature luminescence of air-equilibrated benzophenone/O-propylated p-tert-butylcalix[4]arene solid powdered samples revealed the existence of a novel emission, in contrast with benzophenone/p-tert-butylcalix[4]arene complexes, where only benzophenone emits. This novel emission was identified as phosphorescence of 1-phenyl-1,2-propanedione, which is formed as the result of an hydrogen atom abstraction reaction of the triplet excited benzophenone from the propoxy substituents of the calixarene. Room temperature phosphorescence was obtained in air-equilibrated samples in all propylated hosts. The decay times of the benzophenone emission vary greatly with the degree of propylation, the shortest lifetimes being obtained in the tri- and tetrapropylated calixarenes. Triplet-triplet absorption of benzophenone was detected in all cases, and is the predominant absorption in the p-tert-butylcalix[4]arene case. where an endo-calix complex is formed. Benzophenone ketyl radical formation occurs with the O-propylated p-tert-butylcalix[4]arenes hosts, suggesting a different type of host/guest molecular arrangement. Diffuse reflectance laser flash photolysis and gas chromatography-mass spectrometry techniques provided complementary information, the former about transient species and the latter regarding the final products formed after light absorption. Product analysis and identification clearly show that the two main degradation photoproducts following laser excitation in the propylated substrates are 1-phenyl-1,2-propanedione and 2-hydroxybenzophenone, although several other minor photodegradation products were identified. A detailed mechanistic analysis is proposed. While the solution photochemistry of benzophenone is dominated by the hydrogen abstraction reaction from suitable hydrogen donors, in these solid powdered samples, the alpha-cleavage reaction also plays an important role. This finding occurs even with one single laser pulse which lasts only a few nanoseconds, and is apparently related to the fact that scattered radiation exists, due to multiple internal reflections possibly trapping light within non-absorbing microcrystals in the sample, and is detected until at least 20 micros after the laser pulse. This could explain how photoproducts thus formed could also be excited with only one laser pulse.