A dataset and benchmark for malaria life-cycle classification in thin blood smear images

Malaria microscopy, microscopic examination of stained blood slides to detect parasite Plasmodium, is considered be a gold standard for detecting life-threatening disease malaria. Detecting the plasmodium requires skilled examiner and may take up 10 15 minutes completely go through whole slide. Due lack medical professionals in underdeveloped or resource-deficient regions, many cases misdiagnosed, which results unavoidable complications. We propose complement by creating deep learning-based method automatically (localize) parasites photograph film. To handle unbalanced nature dataset, we adopt two-stage approach. Where first stage trained classify cells into just healthy infected. The second each detected cell further malaria life-cycle stage. facilitate research machine introduce new large-scale image dataset. Thirty-eight thousand are tagged from 345 images different Giemsa-stained samples. Extensive experimentation performed using Convolutional Neural Networks on this Our experiments analysis reveal that approach works better than one-stage detection. ensure usability our approach, have also developed mobile app will used local hospitals investigation educational purposes. its annotations, implementation codes released upon publication paper.