Parallel CNN-ELM: A multiclass classification of chest X-ray images to identify seventeen lung diseases including COVID-19.

Numerous epidemic lung diseases such as COVID-19, tuberculosis (TB), and pneumonia have spread over the world, killing millions of people. Medical specialists have experienced challenges in correctly identifying these diseases due to their subtle differences in Chest X-ray images (CXR). To assist the medical experts, this study proposed a computer-aided lung illness identification method based on the CXR images. For the first time, 17 different forms of lung disorders were considered and the study was divided into six trials with each containing two, two, three, four, fourteen, and seventeen different forms of lung disorders. The proposed framework combined robust feature extraction capabilities of a lightweight parallel convolutional neural network (CNN) with the classification abilities of the extreme learning machine algorithm named CNN-ELM. An optimistic accuracy of 90.92% and an area under the curve (AUC) of 96.93% was achieved when 17 classes were classified side by side. It also accurately identified COVID-19 and TB with 99.37% and 99.98% accuracy, respectively, in 0.996 microseconds for a single image. Additionally, the current results also demonstrated that the framework could outperform the existing state-of-the-art (SOTA) models. On top of that, a secondary conclusion drawn from this study was that the prospective framework retained its effectiveness over a range of real-world environments, including balanced-unbalanced or large-small datasets, large multiclass or simple binary class, and high- or low-resolution images. A prototype Android App was also developed to establish the potential of the framework in real-life implementation.

Explainable Transformer-Based Deep Learning Model for the Detection of Malaria Parasites from Blood Cell Images.

Malaria is a life-threatening disease caused by female anopheles mosquito bites. Various plasmodium parasites spread in the victim's blood cells and keep their life in a critical situation. If not treated at the early stage, malaria can cause even death. Microscopy is a familiar process for diagnosing malaria, collecting the victim's blood samples, and counting the parasite and red blood cells. However, the microscopy process is time-consuming and can produce an erroneous result in some cases. With the recent success of machine learning and deep learning in medical diagnosis, it is quite possible to minimize diagnosis costs and improve overall detection accuracy compared with the traditional microscopy method. This paper proposes a multiheaded attention-based transformer model to diagnose the malaria parasite from blood cell images. To demonstrate the effectiveness of the proposed model, the gradient-weighted class activation map (Grad-CAM) technique was implemented to identify which parts of an image the proposed model paid much more attention to compared with the remaining parts by generating a heatmap image. The proposed model achieved a testing accuracy, precision, recall, f1-score, and AUC score of 96.41%, 96.99%, 95.88%, 96.44%, and 99.11%, respectively, for the original malaria parasite dataset and 99.25%, 99.08%, 99.42%, 99.25%, and 99.99%, respectively, for the modified dataset. Various hyperparameters were also finetuned to obtain optimum results, which were also compared with state-of-the-art (SOTA) methods for malaria parasite detection, and the proposed method outperformed the existing methods.

Applying supervised contrastive learning for the detection of diabetic retinopathy and its severity levels from fundus images.

Diabetic Retinopathy (DR) is a major complication in human eyes among the diabetic patients. Early detection of the DR can save many patients from permanent blindness. Various artificial intelligent based systems have been proposed and they outperform human analysis in accurate detection of the DR. In most of the traditional deep learning models, the cross-entropy is used as a common loss function in a single stage end-to-end training method. However, it has been recently identified that this loss function has some limitations such as poor margin leading to false results, sensitive to noisy data and hyperparameter variations. To overcome these issues, supervised contrastive learning (SCL) has been introduced. In this study, SCL method, a two-stage training method with supervised contrastive loss function was proposed for the first time to the best of authors' knowledge to identify the DR and its severity stages from fundus images (FIs) using "APTOS 2019 Blindness Detection" dataset. "Messidor-2" dataset was also used to conduct experiments for further validating the model's performance. Contrast Limited Adaptive Histogram Equalization (CLAHE) was applied for enhancing the image quality and the pre-trained Xception CNN model was deployed as the encoder with transfer learning. To interpret the SCL of the model, t-SNE method was used to visualize the embedding space (unit hyper sphere) composed of 128 D space into a 2 D space. The proposed model achieved a test accuracy of 98.36%, and AUC score of 98.50% to identify the DR (Binary classification) and a test accuracy of 84.364%, and AUC score of 93.819% for five stages grading with the APTOS 2019 dataset. Other evaluation metrics (precision, recall, F1-score) were also determined with APTOS 2019 as well as with Messidor-2 for analyzing the performance of the proposed model. It was also concluded that the proposed method achieved better performance in detecting the DR compared to the conventional CNN without SCL and other state-of-the-art methods.

Surface plasmon resonance biosensor based on hexagonal lattice dual-core photonic crystal fiber.

In this paper, the guiding properties and sensor performance are numerically investigated for a dual-core hexagonal lattice photonic crystal fiber sensor based on surface plasmon resonance (SPR). Gold is used as the active plasmonic material in order to create resonance, and it is placed outside the fiber structure to facilitate the fabrication process. The finite-element method is used to numerically investigate the characteristics of the sensor. By means of wavelength and amplitude interrogation methods, it is found that the proposed sensor shows maximum wavelength sensitivity of 16,000 nm/RIU with $6.25 \times {10^{ - 6}}\,\,{\rm RIU}$6.25×10-6RIU resolution and amplitude sensitivity of $2255\,\,{{\rm RIU}^{ - 1}}$2255RIU-1 with $4.40 \times {10^{ - 6}}\,\,{\rm RIU}$4.40×10-6RIU resolution. The proposed SPR sensor can detect the analyte refractive index ranging from 1.33 to 1.40. This work also includes an investigation of the effect of changing the gold layer thickness, air-hole diameter, and analyte layer on the sensor performance from the optimized design. The proposed sensor could be employed to detect biological and biochemical analytes because of its simple design and promising results.

Tellurite glass defect-core spiral photonic crystal fiber with low loss and large negative flattened dispersion over S + C + L + U wavelength bands.

A defected-core spiral photonic crystal fiber is proposed to achieve very large negative flattened dispersion and small confinement loss. Simulation results reveal that the designed structure exhibits very large flattened dispersion over S+C+L+U wavelength bands and an average dispersion of about -720.7 ps nm(-1) km(-1) with an absolute dispersion variation of 12.7 ps nm(-1) km(-1) over the wavelength ranging from 1.45 to 1.65 µm. The proposed fiber has five air-hole rings in the cladding leading to very small confinement loss of 0.00111 dB/km at the excitation wavelength of 1.55 µm. The tolerance of the fiber dispersion of ±2% changing in the structural parameters is investigated for practical conditions.