Figure 4 - uploaded by M. Suriya
Content may be subject to copyright.
Source publication
The standard method for examining malarial disease is performed through the examination of blood smears under the microscope for parasite-infected red blood cells and this is done by qualified technicians. The inadequacy of this traditional method is enhanced using advanced computer vision and deep learning methods to automatically classify the mal...
Context in source publication
Similar publications
Background & objective:
Ki-67 evaluation is an essential tool to define luminal A and B breast cancers, which is not yet systematized. The International Ki67 in Breast Cancer Working Group suggests the counting of 500 or 1000 cancer cells, which is a time-consuming process. Therefore, novel methods, such as the Eye-10 method and stepwise counting...
Introduction:
The Finnish Diabetes Risk Score (FINDRISC) is a tool that was initially developed to predict the risk of developing type 2 diabetes mellitus in adults. This tool is simple, quick to apply, non-invasive, and low-cost. The aims of this study were to perform a translation and cultural adaptation of the original version of FINDRISC into B...
Background: Sexual health is a human right, yet there is a lack of comprehensive, standardized and culturally appropriate instruments to assess sexual health in Iran.
Objectives: This study aimed to assess the psychometric properties of the Persian version of the Sexual and Reproductive Health Illustrative Questionnaire (SRH-IQ).
Material and metho...
The new satellites sensors such as GeoEye , WorldView and Ikonos , provide new data for better delineation, detection, and visualization of Land cover changes. Several techniques have been proposed and developed for change detection automatically. The main aim of this research is to assess the use of very high-resolution satellites images in monito...
Background and aim:
Point-of-care ultrasound (POCUS) is a noninvasive alternative to ileocolonoscopy for monitoring disease activity in inflammatory bowel disease (IBD) but is underutilized in practice. Accuracy data are needed to engender clinician confidence in POCUS and increase uptake. The aim of this study was to evaluate the accuracy of POCU...
Citations
... In study 6, an experiment is done with batch sizes of 16, 32, 64, and 128 and the highest accuracy is achieved with a batch size of 16 resulting in 96.93 % test accuracy. The flatten layer converts n-dimensional vector representations into 1-dimensional column vector representations for processing at the dense layer, transforming the feature map into a flattened output [111]. An experiment with different types of flatten layers is shown in study 7 where "Flatten" yields the highest accuracy, 96.93. ...
Introduction
Breast cancer stands as the second most deadly form of cancer among women worldwide. Early diagnosis and treatment can significantly mitigate mortality rates.
Purpose
The study aims to classify breast ultrasound images into benign and malignant tumors. This approach involves segmenting the breast's region of interest (ROI) employing an optimized UNet architecture and classifying the ROIs through an optimized shallow CNN model utilizing an ablation study.
Method
Several image processing techniques are utilized to improve image quality by removing text, artifacts, and speckle noise, and statistical analysis is done to check the enhanced image quality is satisfactory. With the processed dataset, the segmentation of breast tumor ROI is carried out, optimizing the UNet model through an ablation study where the architectural configuration and hyperparameters are altered. After obtaining the tumor ROIs from the fine-tuned UNet model (RKO-UNet), an optimized CNN model is employed to classify the tumor into benign and malignant classes. To enhance the CNN model's performance, an ablation study is conducted, coupled with the integration of an attention unit. The model's performance is further assessed by classifying breast cancer with mammogram images.
Result
The proposed classification model (RKONet-13) results in an accuracy of 98.41 %. The performance of the proposed model is further compared with five transfer learning models for both pre-segmented and post-segmented datasets. K-fold cross-validation is done to assess the proposed RKONet-13 model's performance stability. Furthermore, the performance of the proposed model is compared with previous literature, where the proposed model outperforms existing methods, demonstrating its effectiveness in breast cancer diagnosis. Lastly, the model demonstrates its robustness for breast cancer classification, delivering an exceptional performance of 96.21 % on a mammogram dataset.
Conclusion
The efficacy of this study relies on image pre-processing, segmentation with hybrid attention UNet, and classification with fine-tuned robust CNN model. This comprehensive approach aims to determine an effective technique for detecting breast cancer within ultrasound images.
... As such researchers have made use of this model to study images from medical and public health studies. For instance, [22] and [23] (Balaram et al., 2022) proposed a CNN model that detects and classifies malaria parasites from blood smears. CNN models have also been applied in determining the presence of CoVID-19 before reaching a massscale level in patients using CT-scan images [24]. ...
... ReLU converts every negative number from the pooling layer to zero [47]. In the flatten layer, 2dimensional feature maps produced in the previous layer are converted into a 1-dimensional feature map to be suitable for the following fully connected layers [22]. The last layer of the convolutional layer has an LSTM layer which overlooks insignificant parts of the preceding layers and carefully updates the important feature as output that is required. ...
... This approach solves the vanishing gradient in the architecture [48,49]. The fully connected layer has several dense layers with the last layer having a softmax activation algorithm; a multiclass classifier for calculating the probabilities to which the five variants belong [22]. Adaptive moment estimation optimizer (Adam) is used as an optimizer in the neural network architecture. ...
High-throughput sequencing techniques and sequence analysis have enabled the taxonomic classification of pathogens present in clinical samples. Sequencing provides an unbiased identification and systematic classification of pathogens and this is generally achieved by comparing novel sequences to pre-existing annotated reference databases. However, this approach is limited by large-scale reference databases which require considerable computational resources and skills to compare against. Alternative robust methods such as machine learning are currently employed in genome sequence analysis and classification, and it can be applied in classifying SARS-CoV-2 variants, whose continued evolution has resulted in the emergence of multiple variants.
We developed a deep learning Convolutional Neural Networks-Long Short Term Memory (CNN-LSTM) model to classify dominant SARS-CoV-2 variants (omicron, delta, beta, gamma and alpha) based on gene sequences from the surface glycoprotein (spike gene). We trained and validated the model using > 26,000 SARS-CoV-2 sequences from the GISAID database. The model was evaluated using unseen 3,057 SARS-CoV-2 sequences. The model was compared to existing molecular epidemiology tool, nextclade.
Our model achieved an accuracy of 98.55% on training, 99.19% on the validation and 98.41% on the test dataset. Comparing the proposed model to nextclade, the model achieved significant accuracy in classifying SARS-CoV-2 variants from unseen data. Nextclade identified the presence of recombinant strains in the evaluation data, a mechanism that the proposed model did not detect.
This study provides an alternative approach to pre-existing methods employed in the classification of SARS-CoV-2 variants. Timely classification will enable effective monitoring and tracking of SARS-CoV-2 variants and inform public health policies in the control and management of the COVID-19 pandemic.
... The applied dataset includes a malaria cell image dataset [8], [23], containing 27,558 image data, divided into two classifications: 13,775 image data for the parasitized class and 13,813 image data for the uninfected class. The sample infected and uninfected images depict in Fig. 1 and Fig. 2. Parasitized class refers to image data of blood cells infected with malaria, while uninfected refers to image data that is not infected with malaria. ...
... Thus, the adam optimizer is expected to provide better accuracy. In addition, the adam optimizer does not require a lot of storage space, and this optimizer is also lighter during the epoch training process [23]. The results of accuracy and loss of scenario 2 are depicted in in Figs. 5 and 6. ...
Malaria is a severe global public health problem caused by the bite of infected mosquitoes. It can be cured, but only with early detection and effective, quick treatment. It can cause severe conditions if not properly diagnosed and treated at an early stage. In the worst scenario, it can cause death. This study aims at focusing on classifying malaria cell images. Malaria is classified as a dangerous disease caused by the bite of the female Anophles mosquito. As such, it leads to mortality when immediate action and treatment fails to be administered. In particular, this study aims to classify malaria cell images by utilizing the Inception-V3 architecture. In this study, training was conducted on 27,558 malaria cell image data through Inception-V3 architecture by proposing 3 scenarios. The proposed scenario 1 model applies the SGD optimizer to generate a loss value of 0.13 and an accuracy value of 0.95; scenario 2 model applies the Adam optimizer to generate a loss value of 0.09 and an accuracy value of 0.96; and lastly scenario 3 implements the RMSprop optimizer to generate a loss value of 0.08 and an accuracy value of 0.97. Applying the three scenarios, the results of the study apparently indicate that the Inception-V3 model using the RMSprop optimizer is capable of providing the best accuracy results with an accuracy of 97% with the lowest loss value, compared to scenario 1 and scenario 2. Further, the test results confirms that the proposed model in this study is capable of classifying malaria cells effectively.
... In disease diagnosis, certain medical terms such as image processing, identification, classification, and segmentation played a critical role in advancement of medical technology and automation, which can help to ease out the pressure on the pathologists [2,3]. Due to advancements in image processing methods and computing power, deep learningbased cervical cancer analysis has become more widely used [4]. One of the deep learning algorithms, CNN, has been widely used to identify and classify cervical cancer. ...
The most prevalent deadly occurring disease nowadays in women is nothing but cervical cancer which has to be avoided by frequently having screenings to detect and cure pre-cancers. Certain tests called Pap-Test well examine the cells of the cervix for any abnormal or precancerous alterations. However, manual microscopic Pap smear screening is instinctive and although necessary but quite difficult process. Thus, the objective of this research was to create a computerized cervical cancer prediction system based on digital image processing of Pap smear pictures which uses a better pooling technique. The novelty of this work is the use of a hybrid pooling technique, which effectively proves to better the results. In the cervical cancer detection system, image analysis of Pap smears is critical. With emerging technologies, computer vision convolutional neural networks (CNNs) are of prior importance because of their incredible potential to cope with proper object detection and image identification problems. In CNN, to reduce the dimensionality of processed input, a pooling method is used which helps to reduce computational costs while also improving noise tolerance and translation tolerance. While basic pooling techniques such as Max pooling and average pooling are commonly used in numerous research, a recently developed pooling strategy should help CNN generalize better. The proposed model is a hybrid pooling approach that picks the maximum or picks average pooling for each pooling layer based on a stochastic decision. The likelihood of choosing one of the two pooling algorithms for each convolutional layer can be modified, which is a feature of hybrid pooling. However, proposed hybrid pooling technique is helpful for boosting the CNNs’ generalization capabilities in image classification tasks utilizing standard datasets. The proposed model, which incorporates the AlexNet-SVM network with a hybrid pooling strategy, has an average accuracy of 95.45% with a very minimal computational time. The average accuracy of the model trained with Max pooling is 93.56%, whereas the model trained with average pooling is 85.71%.
... DL has also shown enormous success in the automated diagnosis of malaria. A 19-layer CNN was demonstrated for classifying infected and uninfected malarial cells, with an accuracy rate of 98.9% [24]. The use of multi-wavelength imaging to aid in data augmentation was used in a quick and robust malaria classification system [25]. ...
Malaria is a lethal disease responsible for thousands of deaths worldwide every year. Manual methods of malaria diagnosis are time-consuming that require a great deal of human expertise and efforts. Computer-based automated diagnosis of diseases is progressively becoming popular. Although deep learning models show high performance in the medical field, it demands a large volume of data for training which is hard to acquire for medical problems. Similarly, labeling of medical images can be done with the help of medical experts only. Several recent studies have utilized deep learning models to develop efficient malaria diagnostic system, which showed promising results. However, the most common problem with these models is that they need a large amount of data for training. This paper presents a computer-aided malaria diagnosis system that combines a semi-supervised generative adversarial network and transfer learning. The proposed model is trained in a semi-supervised manner and requires less training data than conventional deep learning models. Performance of the proposed model is evaluated on a publicly available dataset of blood smear images (with malaria-infected and normal class) and achieved a classification accuracy of 96.6%.
... e purpose of medical image processing is to restore the original unclear image, to highlight some characteristic information in the image, or to classify the image. Medical images include MRI, CT, ultrasound images, and blood smear images [11,12]. Convolutional neural network (CNN) is an important end-toend deep learning model [13], which is mainly used in image recognition, segmentation, and target detection in medical image processing. ...
With the rapid development of deep learning, automatic lesion detection is used widely in clinical screening. To solve the problem that existing deep learning-based cervical precancerous lesion detection algorithms cannot meet high classification accuracy and fast running speed at the same time, a ShuffleNet-based cervical precancerous lesion classification method is proposed. By adding channel attention to the ShuffleNet, the network performance is improved. In this study, the image dataset is classified into five categories: normal, cervical cancer, LSIL (CIN1), HSIL (CIN2/CIN3), and cervical neoplasm. The colposcopy images are expanded to solve the problems of the lack of colposcopy images and the uneven distribution of images from each category. For the test dataset, the accuracy of the proposed CNN models is 81.23% and 81.38%. Our classifier achieved an AUC score of 0.99. The experimental results show that the colposcopy image classification network based on artificial intelligence has good performance in classification accuracy and model size, and it has high clinical applicability.
... Therefore, to answer how to provide innovation with a large medical dataset in helping malaria elimination efforts, this study aims to identify the dataset in the form of images of malaria infected and non-malaria (normal cells and leukemia cancer-mutated cells) as a means of malaria diagnosis, using a Deep Learning model algorithm which is Convolutional Neural Network (CNN). As a deep learning model, CNN is specifically designed to study two-dimensional (2D) data such as images and videos (Liang et al., 2017) that can be used in the recognition and classification process (Suriya et al., 2019). By conducting this research, it is hoped that in its future development it can help in decision making and efforts to eliminate malaria, especially in Eastern Indonesia. ...
... Presenting three new stages in the diagnosis process of malaria, this research used red blood cells to apply the Segmentation Neural Network (SNN) for image segmentation with 93.72% accuracy and the CNN method for image classification 87.04% (Delgado-Ortet et al., 2020). Unlike the case with Suriya et all., who proposed a Deep Convolutional Neural Network that focuses on comparing validation loss and accuracy by setting hyper-parameters to classify images and calculating the Kappa coefficient and Matthew's correlation coefficient (Suriya et al., 2019). The accuracy obtained is 98.9%. ...
Malaria is a contagious infectious disease that is still threatening human life. Malaria morbidity when viewed by province shows that Eastern Indonesia is the area with the highest Annual Parasite Incidence (API), namely Papua, West Papua, NTT, and Maluku. This is a concern for the continued efforts to control and eliminate malaria in these high malaria-endemic areas. There are many strategies to help and prevent, include the possibility of innovation in the diagnostic process. Therefore, to answer how to provide innovation in technology to accelerate the elimination of malaria, this study aims to identify the image of red blood cells which infected with malaria among other normal and leukemia cancer-mutated cells (non-malaria) by making improvements through the proposed new model used. This model is meant to do deep learning using Convolutional Neural Network (CNN). The results obtained in this study show that the success of using the proposed model is influenced by the pre-processing stage, the dropout regularization function, learning rate, and momentum value used. The accuracy value obtained is 0.9660, 0.9693 precision, 0.9626 recall, and an F1 score of 0.9659.
... In that research, only infected and non-infected malaria parasite microscopic blood cell images were targeted. Another researcher Suriya et al. [35] developed a deep neural network architecture trained using fine-tuning and settings of hyperparameters for the classification of malaria-infected and non-infected cells. Tehreem et al. [36] proposed adaptive thresholding and morphological-based method using bilateral filtering to remove the noise and enhance the image quality based on image processing algorithms to detect malaria parasite inside each cell. ...
Background and objectives
Microscopic images are an important part for haematologists in diagnosing various diseases in the blood cell. Changes in blood cells are caused by malaria disease, and early diagnosis can prevent the disease from entering its severe stage.
Methods
In this paper, an automated non-invasive and efficient deep learning-based framework is developed for multi-class plasmodium vivax life cycle classification and malaria diagnosis. A multi-class microscopic blood cell of different plasmodium vivax life cycle stage dataset is analysed, and a diagnostic framework is designed. Several stages of the disease are examined and augmented through various techniques to make the framework robust in real-time. Generative adversarial network is specially designed to generate extended training samples of various life cycle stages to increase robustness of the resulting model. A special transformer-based neural network vision transformer is designed to improve generalisation capabilities. Microscopic images are classified into multi classes of plasmodium vivax life cycle stage, where the keystone transformer layers extract relevant disease features from microscopic colour images, and the extracted relevant features are used to make predictive diagnostic decisions.
Results
The capabilities of the vision transformer are computed and analysed by statistical parameters, and the performance of the vision transformer model is compared with baseline architectures, where it was evident that the performance of the vision transformer was significantly better, reaching 90.03% accuracy.
Conclusions
A comprehensive comparison of the proposed framework to the state-of-the-art methods proves its efficiency in the classification of plasmodium vivax life cycle for malaria disease identification through thin blood smear microscopic images.
... e evolution of DL and CV in medicine has solved most medical imaging and other medical-related problems such as Alzheimer detection, cervical cancer, malarial detection [12], and brain tumour [13]. Infection is defined as at least two classic signs of purulence in DFU. ...
A bacterial or bone infection in the feet causes diabetic foot infection (DFI), which results in reddish skin in the wound and surrounding area. DFI is the most prevalent and dangerous type of diabetic mellitus. It will mainly occur in people with heart disease, renal illness, or eye disease. The clinical signs and symptoms of local inflammation are used to diagnose diabetic foot infection. In assessing diabetic foot ulcers, the infection has significant clinical implications in predicting the likelihood of amputation. In this work, a diabetic foot infection network (DFINET) is proposed to assess infection and no infection from diabetic foot ulcer images. A DFINET consists of 22 layers with a unique parallel convolution layer with ReLU, a normalization layer, and a fully connected layer with a dropout connection. Experiments have shown that the DFINET, when combined with this technique and improved image augmentation, should yield promising results in infection recognition, with an accuracy of 91.98%, and a Matthews correlation coefficient of 0.84 on binary classification. Such enhancements to existing methods shows that the suggested approach can assist medical experts in automated detection of DFI.
... Por ejemplo, en (Shorten C. et. al, 2021;Suriva et al, 2019;Shah et. al, 2020) se reportan trabajos donde se realizan clasificadores binarios utilizando redes neuronales convolucionales (CNN). ...
... vamente), 5 capas de max-pooling y 2 capas totalmente conectadas. El conjunto de imágenes utilizado es del conjunto de datos NIH Malaria, donde se tienen 13779 imágenes con parásitos y 13779 sin parásitos. Este conjunto de datos fue aumentado mediante técnicas clásicas hasta obtener 173700 imágenes. El clasificador tuvo una precisión de 96.33%. En (Suriva . et al, 2019) se realiza una clasificación de imágenes de células rojas infectadas debido a presencia de parásitos. Primeramente, la imagen es procesada para tener un tamaño de 128x128. Luego se entrena la red diseñada. La cual tiene 19 capas de convolución para la extracción de características, 6 capas de max pooling para reducir el número de cálcul ...
Los seres humanos pueden albergar parásitos, por lo tanto, es fundamental una detección temprana para prevenir enfermedades. Los parásitos pueden observarse a través de imágenes microscópicas, con lo cual, la visión por computadora se muestra como un enfoque que pueda ayudar a la detección y clasificación de parásitos en imágenes digitales. Los modelos de aprendizaje profundo han mostrado un desempeño formidable en la clasificación de imágenes, debido a esto, en este artículo se presentan varios clasificadores profundos multiclase para reconocer 8 clases: 7 tipos de parásitos y la clase no parásitos. Los clasificadores diseñados utilizan transferencia de aprendizaje basada en la arquitectura AlexNet modificada. Al contar con una cantidad de muestras escasa de imágenes de parásitos, se realizó también un aumento de datos, tanto con el método tradicional como mediante la generación de imágenes con una red generativa adversaria (GAN) diseñada para ese propósito. El clasificador con mejor desempeño presentó una exactitud del 99.94%, 98.97% y 98.18% en el conjunto de entrenamiento, validación y de prueba, respectivamente.
