Decolonizing global AI governance: assessment of the state of decolonized AI governance in Sub-Saharan Africa.

Global artificial intelligence (AI) governance must prioritize equity, embrace a decolonial mindset, and provide the Global South countries the authority to spearhead solution creation. Decolonization is crucial for dismantling Western-centric cognitive frameworks and mitigating biases. Integrating a decolonial approach to AI governance involves recognizing persistent colonial repercussions, leading to biases in AI solutions and disparities in AI access based on gender, race, geography, income and societal factors. This paradigm shift necessitates deliberate efforts to deconstruct imperial structures governing knowledge production, perpetuating global unequal resource access and biases. This research evaluates Sub-Saharan African progress in AI governance decolonization, focusing on indicators like AI governance institutions, national strategies, sovereignty prioritization, data protection regulations, and adherence to local data usage requirements. Results show limited progress, with only Rwanda notably responsive to decolonization among the ten countries evaluated; 80% are 'decolonization-aware', and one is 'decolonization-blind'. The paper provides a detailed analysis of each nation, offering recommendations for fostering decolonization, including stakeholder involvement, addressing inequalities, promoting ethical AI, supporting local innovation, building regional partnerships, capacity building, public awareness, and inclusive governance. This paper contributes to elucidating the challenges and opportunities associated with decolonization in SSA countries, thereby enriching the ongoing discourse on global AI governance.

Structured adaptive boosting trees for detection of multicellular aggregates in fluorescence intravital microscopy.

Fluorescence intravital microscopy captures large data sets of dynamic multicellular interactions within various organs such as the lungs, liver, and brain of living subjects. In medical imaging, edge detection is used to accurately identify and delineate important structures and boundaries inside the images. To improve edge sharpness, edge detection frequently requires the inclusion of low-level features. Herein, a machine learning approach is needed to automate the edge detection of multicellular aggregates of distinctly labeled blood cells within the microcirculation. In this work, the Structured Adaptive Boosting Trees algorithm (AdaBoost.S) is proposed as a contribution to overcome some of the edge detection challenges related to medical images. Algorithm design is based on the observation that edges over an image mask often exhibit special structures and are interdependent. Such structures can be predicted using the features extracted from a bigger image patch that covers the image edge mask. The proposed AdaBoost.S is applied to detect multicellular aggregates within blood vessels from the fluorescence lung intravital images of mice exposed to e-cigarette vapor. The predictive capabilities of this approach for detecting platelet-neutrophil aggregates within the lung blood vessels are evaluated against three conventional machine learning algorithms: Random Forest, XGBoost and Decision Tree. AdaBoost.S exhibits a mean recall, F-score, and precision of 0.81, 0.79, and 0.78, respectively. Compared to all three existing algorithms, AdaBoost.S has statistically better performance for recall and F-score. Although AdaBoost.S does not outperform Random Forest in precision, it remains superior to the XGBoost and Decision Tree algorithms. The proposed AdaBoost.S is widely applicable to analysis of other fluorescence intravital microscopy applications including cancer, infection, and cardiovascular disease.

DeepLeish: a deep learning based support system for the detection of Leishmaniasis parasite from Giemsa-stained microscope images.

BACKGROUND: Leishmaniasis is a vector-born neglected parasitic disease belonging to the genus Leishmania. Out of the 30 Leishmania species, 21 species cause human infection that affect the skin and the internal organs. Around, 700,000 to 1,000,000 of the newly infected cases and 26,000 to 65,000 deaths are reported worldwide annually. The disease exhibits three clinical presentations, namely, the cutaneous, muco-cutaneous and visceral Leishmaniasis which affects the skin, mucosal membrane and the internal organs, respectively. The relapsing behavior of the disease limits its diagnosis and treatment efficiency. The common diagnostic approaches follow subjective, error-prone, repetitive processes. Despite, an ever pressing need for an accurate detection of Leishmaniasis, the research conducted so far is scarce. In this regard, the main aim of the current research is to develop an artificial intelligence based detection tool for the Leishmaniasis from the Geimsa-stained microscopic images using deep learning method. METHODS: Stained microscopic images were acquired locally and labeled by experts. The images were augmented using different methods to prevent overfitting and improve the generalizability of the system. Fine-tuned Faster RCNN, SSD, and YOLOV5 models were used for object detection. Mean average precision (MAP), precision, and Recall were calculated to evaluate and compare the performance of the models. RESULTS: The fine-tuned YOLOV5 outperformed the other models such as Faster RCNN and SSD, with the MAP scores, of 73%, 54% and 57%, respectively. CONCLUSION: The currently developed YOLOV5 model can be tested in the clinics to assist the laboratorists in diagnosing Leishmaniasis from the microscopic images. Particularly, in low-resourced healthcare facilities, with fewer qualified medical professionals or hematologists, our AI support system can assist in reducing the diagnosing time, workload, and misdiagnosis. Furthermore, the dataset collected by us will be shared with other researchers who seek to improve upon the detection system of the parasite. The current model detects the parasites even in the presence of the monocyte cells, but sometimes, the accuracy decreases due to the differences in the sizes of the parasite cells alongside the blood cells. The incorporation of cascaded networks in future and the quantification of the parasite load, shall overcome the limitations of the currently developed system.

Vision-Transformer-Based Transfer Learning for Mammogram Classification.

Breast mass identification is a crucial procedure during mammogram-based early breast cancer diagnosis. However, it is difficult to determine whether a breast lump is benign or cancerous at early stages. Convolutional neural networks (CNNs) have been used to solve this problem and have provided useful advancements. However, CNNs focus only on a certain portion of the mammogram while ignoring the remaining and present computational complexity because of multiple convolutions. Recently, vision transformers have been developed as a technique to overcome such limitations of CNNs, ensuring better or comparable performance in natural image classification. However, the utility of this technique has not been thoroughly investigated in the medical image domain. In this study, we developed a transfer learning technique based on vision transformers to classify breast mass mammograms. The area under the receiver operating curve of the new model was estimated as 1 ± 0, thus outperforming the CNN-based transfer-learning models and vision transformer models trained from scratch. The technique can, hence, be applied in a clinical setting, to improve the early diagnosis of breast cancer.

Squamous Cell Carcinoma of Skin Cancer Margin Classification From Digital Histopathology Images Using Deep Learning.

OBJECTIVES: Now a days, squamous cell carcinoma (SCC) margin assessment is done by examining histopathology images and inspection of whole slide images (WSI) using a conventional microscope. This is time-consuming, tedious, and depends on experts' experience which may lead to misdiagnosis and mistreatment plans. This study aims to develop a system for the automatic diagnosis of skin cancer margin for squamous cell carcinoma from histopathology microscopic images by applying deep learning techniques. METHODS: The system was trained, validated, and tested using histopathology images of SCC cancer locally acquired from Jimma Medical Center Pathology Department from seven different skin sites using an Olympus digital microscope. All images were preprocessed and trained with transfer learning pre-trained models by fine-tuning the hyper-parameter of the selected models. RESULTS: The overall best training accuracy of the models become 95.3%, 97.1%, 89.8%, and 89.9% on EffecientNetB0, MobileNetv2, ResNet50, VGG16 respectively. In addition to this, the best validation accuracy of the models was 94.7%, 91.8%, 87.8%, and 86.7% respectively. The best testing accuracy of the models at the same epoch was 95.2%, 91.5%, 87%, and 85.5% respectively. From these models, EfficientNetB0 showed the best average training and testing accuracy than the other models. CONCLUSIONS: The system assists the pathologist during the margin assessment of SCC by decreasing the diagnosis time from an average of 25 minutes to less than a minute.

Developing a desktop application for drug-drug interaction checker ordered for chronic diseases in Ethiopian hospitals pharmacy.

BACKGROUND: Drug-drug interactions are a major cause of morbidity worldwide and are a leading source of treatment inefficacy. They are classified based on their pharmacological action on the body as major, moderate, and minor. Currently, it is a tedious process to remember the drug-drug interactions by the pharmacist during dispensing of the prescribed drugs for the patients. Therefore, there is a need for technology that assists the pharmacist in checking the drug-drug interaction for prescribed drugs. Therefore, in this work, a desktop-based application that can automatically identify the drug-drug interactions for prescribed drugs that could operate offline for those found in low-resource setting hospitals has been developed. To do this, around 3000 drugs along with their major and moderate interaction points were collected from Ethiopian Pharmaceutical Supply Agency. The developed system included two main parts; the database part that comprises all the drugs collected along with their major and moderate interaction points, and a patient registration platform to register the patients' history. The system was developed by using C sharp programing language. RESULTS: The developed system has both drug-drug interaction checking as well as patient registration platform. Registration of the patient's history will be done by the pharmacist and during dispensing of the drugs to the patient, the developed system will check the interaction between the drugs prescribed. The system was tested to operate the above functions, and finally, it was able to display the major and moderate interaction points of all inserted drugs automatically and accurately. For those drugs which have no either major or moderate interaction, the system was displayed as 'unknown'. CONCLUSIONS: The developed system assisted the pharmacist in knowing the drug-drug interaction, and enabled the patients for the resubscription of drugs with the same functional. The system would help to increase the efficiency of the pharmacist in low resource settings to do their tasks without any difficulty, and tiredness. In the future, it is recommended to include all drugs for all disease types rather than focusing only on chronic disease drugs.

Low cost, non-invasive, and continuous vital signs monitoring device for pregnant women in low resource settings (Lvital device).

Around 800 women die each day from complications of pregnancy and childbirth in the world. Vital Signs monitoring (such as blood pressure, pulse rate, and temperature) are among fundamental parameters of ensuring the health and safety of women and newborns during pregnancy, labor, and childbirth. Approximately, 10% of women experience hypertension (greater than140/90) during pregnancy. High blood pressure during pregnancy can cause extra stress on the heart and kidneys and can increase the risk of heart disease. Therefore, early recognition of abnormal vital signs, which are induced due to pregnancy can allow for timely identification of clinical deterioration. Currently used technologies are expensive and complex design with implementation challenges in low-resource settings where maternal morbidity and mortality is higher. Thus, considering the above need, here a hardware device has been designed and developed, which is a low cost, and portable for pregnant women's vital signs (with cuff-less blood pressure, heart rate, and body temperature) monitoring device. The developed device would have a remarkable benefit of monitoring the maternal vital signs especially for those in low resource settings, where there is a high paucity of vital signs monitoring devices.

De-Speckling Breast Cancer Ultrasound Images Using a Rotationally Invariant Block Matching Based Non-Local Means (RIBM-NLM) Method.

The ultrasonic technique is an indispensable imaging modality for diagnosis of breast cancer in young women due to its ability in efficiently capturing the tissue properties, and decreasing nega-tive recognition rate thereby avoiding non-essential biopsies. Despite the advantages, ultrasound images are affected by speckle noise, generating fine-false structures that decrease the contrast of the images and diminish the actual boundaries of tissues on ultrasound image. Moreover, speckle noise negatively impacts the subsequent stages in image processing pipeline, such as edge detec-tion, segmentation, feature extraction, and classification. Previous studies have formulated vari-ous speckle reduction methods in ultrasound images; however, these methods suffer from being unable to retain finer edge details and require more processing time. In this study, we propose a breast ultrasound de-speckling method based on rotational invariant block matching non-local means (RIBM-NLM) filtering. The effectiveness of our method has been demonstrated by com-paring our results with three established de-speckling techniques, the switching bilateral filter (SBF), the non-local means filter (NLMF), and the optimized non-local means filter (ONLMF) on 250 images from public dataset and 6 images from private dataset. Evaluation metrics, including Self-Similarity Index Measure (SSIM), Peak Signal to Noise Ratio (PSNR), and Mean Square Error (MSE) were utilized to measure performance. With the proposed method, we were able to record average SSIM of 0.8915, PSNR of 65.97, MSE of 0.014, RMSE of 0.119, and computational speed of 82 seconds at noise variance of 20dB using the public dataset, all with p-value of less than 0.001 compared against NLMF, ONLMF, and SBF. Similarly, the proposed method achieved av-erage SSIM of 0.83, PSNR of 66.26, MSE of 0.015, RMSE of 0.124, and computational speed of 83 seconds at noise variance of 20dB using the private dataset, all with p-value of less than 0.001 compared against NLMF, ONLMF, and SBF.

Treatment device for neonatal birth asphyxia related Hypoxic Ischemic Encephalopathy.

BACKGROUND: Birth asphyxia is a leading cause of neonatal brain injury, morbidity, and mortality globally. It leads to a multi-organ dysfunction in the neonate and to a neurological dysfunction called Hypoxic Ischemic Encephalopathy (HIE). Cooling therapy is commonly used to slow or stop the damaging effects of birth asphyxia. However, most of the cooling devices used in the healthcare facility do not have a rewarming functionality after cooling therapy. A separate rewarming device, usually a radiant warmer or incubator is used to rewarm the infant after therapy, causing additional burden to the healthcare system and infant families. The objective of this project was, therefore, to design and develop a cost-effective and efficient total body cooling and rewarming device. METHODS: Our design includes two water reservoirs that operate by pumping cold and warm sterile water to a mattress. After decreasing the infant's core body temperature to 33.5 °C, the system is designed to maintain it for 72 h. Feedback for temperature regulation is provided by the rectal and mattress temperature sensors. Once the cooling therapy is completed, the system again rewarms the water inside the mattress and gradually increases the neonate temperature to 36.5-37 °C. The water temperature sensors' effectiveness was evaluated by adding 1000 ml of water to the reservoir and cooling and warming to the required level of temperature using Peltier. Then a digital thermometer was used as a gold standard to compare with the sensor's readings. This was performed for five iterations. RESULTS: The prototype was built and gone through different tests and iterations. The proposed device was tested for accuracy, cost-effectiveness and easy to use. Ninety-three point two percent accuracy has been achieved for temperature sensor measurement, and the prototype was built only with a component cost of less than 200 USD. This is excluding design, manufacturing, and other costs. CONCLUSION: A device that can monitor and regulate the neonate core body temperature at the neuroprotective range is designed and developed. This is achieved by continuous monitoring and regulation of the water reservoirs, mattress, and rectal temperatures. The device also allows continuous monitoring of the infant's body temperature, mattress temperature, reservoir temperature, and pulse rate. The proposed device has the potential to play a significant role in reducing neonatal brain injury and death due to HIE, especially in low resource settings, where the expertise and the means are scarce.

Accurate Machine-Learning-Based classification of Leukemia from Blood Smear Images.

BACKGROUND: Conventional identification of blood disorders based on visual inspection of blood smears through microscope is time consuming, error-prone and is limited by hematologist's physical acuity. Therefore, an automated optical image processing system is required to support the clinical decision-making. MATERIALS AND METHODS: Blood smear slides (n = 250) were prepared from clinical samples, imaged and analyzed in Jimma Medical Center, Hematology department. Samples were collected, analyzed and preserved from out and in-patients. The system was able to categorize four common types of leukemia's such as acute and chronic myeloid leukemia; and acute and chronic lymphoblastic leukemia, through a robust image segmentation protocol, followed by classification using the support vector machine. RESULTS: The system was able to classify leukemia types with an accuracy, sensitivity, specificity of 97.69%, 97.86% and 100%, respectively for the test datasets, and 97.5%, 98.55% and 100%, respectively, for the validation datasets. In addition, the system also showed an accuracy of 94.75% for the WBC counts that include both lymphocytes and monocytes. The computer-assisted diagnosis system took less than one minute for processing and assigning the leukemia types, compared to an average period of 30 minutes by unassisted manual approaches. Moreover, the automated system complements the healthcare workers' in their efforts, by improving the accuracy rates in diagnosis from ∼70% to over 97%. CONCLUSION: Importantly, our module is designed to assist the healthcare facilities in the rural areas of sub-Saharan Africa, equipped with fewer experienced medical experts, especially in screening patients for blood associated diseases including leukemia.

Transfer Learning in Breast Cancer Diagnoses via Ultrasound Imaging.

Transfer learning is a machine learning approach that reuses a learning method developed for a task as the starting point for a model on a target task. The goal of transfer learning is to improve performance of target learners by transferring the knowledge contained in other (but related) source domains. As a result, the need for large numbers of target-domain data is lowered for constructing target learners. Due to this immense property, transfer learning techniques are frequently used in ultrasound breast cancer image analyses. In this review, we focus on transfer learning methods applied on ultrasound breast image classification and detection from the perspective of transfer learning approaches, pre-processing, pre-training models, and convolutional neural network (CNN) models. Finally, comparison of different works is carried out, and challenges-as well as outlooks-are discussed.