Diagnosing Malaria: Artificial Intelligence for Global Health

Jun 06,2019 by Gauri Mishra
Inner banner

Malaria, till date remains a deadly disease globally. It has been estimated that nearly 200 million cases of malaria occur every year worldwide and roughly 400,000 deaths take place due to this contagious disease.

Apart from medical research, modern software technology is having a chief role in diagnosing infectious diseases. To be more particular one of the impediments for mortality mitigation has been inaccurate malaria detection.

Thus, to enhance detection of malaria, image analysis software, as well as artificial intelligence based strategies, are getting employed to evaluate parasitemia under microscopic blood slides. This blog provides an outline of such procedures and explains the present advancements in image analysis and artificial intelligence for automated microscopic malaria diagnosis.

Application of Artificial Intelligence is contributing to malaria detection with numerous methods such as feature computation, imaging, parasite detection, image preprocessing, cell segmentation and automatic cell classification. 

Malaria is considered as an extremely contagious disease globally which cause loads of health issues in several tropical countries. Plasmodium falciparum, which is known as a vital human malaria parasite, is considered to cause critical illnesses followed by thousands of deaths every year. 

Recently at Indian Institute of Technology-Delhi, researchers have reported that they have designed an Artificial Intelligence-driven digital hardware system to diagnose malaria, intestinal parasite, tuberculosis as well as cervical cancer just within some milliseconds. They have built an artificial intelligence oriented less-power electronic hardware systems which will assist in diagnosing various infectious diseases instantly.

No matter how much attempts we do to manage malaria, detecting it in all the spheres of the globe yet needs calculating malaria parasites under the automated microscope over a glass slide smudged with blood. This the era, where an artificial intelligence based code will do diagnosis more effectively in comparison to humans.

The motivation of this article is all about understanding the significance of AI in the deadly disease malaria and the productiveness of software in diagnosing malaria.

  • AI accelerating Medical Research
  • Convolutional Neural Networks (CNN)

Artificial Intelligence is the talking trend in Information Technology that has enhanced the performance in several medical fields. With daily physical detection of blood smudges, it is a thorough but time taking procedure needing perfect experience in organizing and evaluating the parasitized as well as uninfected cells. Generally, it might not count properly and can cause issues if we don’t possess the perfect practice in particular locations globally.

Various enhancements are done in leveraging progressive image processing as well as analysis procedures to remove physically-designed options and design machine learning driven detecting frameworks. Although such frameworks aren’t measurable with additional info provided for practice and the manual methods consume loads of time. Deep Learning or AI-based models, or in fact to be particular, Convolutional Neural Networks (CNN) have demonstrated to be highly productive in a broad range of computer vision activities. 

Those convolution layers study ‘spatial hierarchical patterns’ given in the information that is even translation constant. Therefore they are capable to go through various facets of images. For example, the primary convolution layer can determine medium and localized patterns like edges along with corners, a secondary convolution layer can understand bigger patterns depending on the options from the primary layers. This enables CNN’s to automatize designing and processing productive options that oversimplify the latest information points. Pooling layers assist with downsampling as well as dimension minimization.

Therefore, CNNs facilitate medical science with automatizing and extendable ‘feature engineering’. In fact, operating within wide layers allows us to conduct operations such as image classification. Automated malaria diagnosis utilizing AI-driven structures like CNNs will be highly productive, affordable and ascendable specifically with the arrival of transfer learning as well as pre-trained frameworks that operate pretty well with various constraints such as less information.

  • A detailed Breakdown of Image Recognition for Malaria

The primary step is generally the procurement of digitalized images of blood stains that is based greatly over the tools as well as materials getting utilized. The Image acquisition segment knocks various approaches for several kinds of microscopy, staining as well as blood slides (thin or thick). With image acquisition, multiple systems carry out one or more preprocessing procedures to eliminate noise and regularize lighting along with color differences intrinsic within the image acquisition and the staining procedure. The Preprocessing segment classifies the publications in accordance to the preprocessing strategies executed.

The next step generally entails the detection and segmentation of each and every blood cells which are visible under a blood slide image, like parasites or platelets. Red blood cell detection and segmentation provides an outline of all the segmentation procedures which are utilized for microscopic malaria detection.

In the final step, a mathematical discrimination procedure which sorts the segmented cells into various classes depending over the computed options is executed. For an instance, labeling an individual red blood cell either as infected or say uninfected is a prime segmenting operation conducted that enables to compute the parasitemia.

·Mobile Smartphones for Malaria Diagnosis

Some researchers have designed a mobile based application for malaria diagnosis which executes over an Android smartphone connected to an old light microscope. The perfect hardware solution for the purpose of microscopic malaria detection in resource-based settings will be a medium ‘portable slide reader’ within which a blood slide will be pushed and provide results of parasitemia.

However information technology is leading this way and to be specific the comparatively extreme optical magnification required (till 1000×) for malaria detection in amalgamation with oil immersion is a pivotal downsizing hindrance, till options are available.

Artificial Intelligence provides tiny camera-built computing tools, like smartphones, that will be connected to a magnifying tool and will then calculate the parasitemia, utilizing image analysis as well as machine learning. Digital era based smartphones are highly robust computing devices and the cameras offer enough resolution for malaria detection.

Furthermore to this, Android devices are comparatively affordable and are generally within the possession of medical workers. However cellular network connectivity will assist with the data exchange through healthcare workers as well as hospitals. Tiny magnifying tools which will be connected to a smartphone’s camera, enabling precise optical magnification in comparison with digitalized zooming, are provided commercially.

Making an Impact

As aforementioned, we went through a real-world medical imaging strategy of malaria diagnosis in this blog. Artificial Intelligence in India for medical treatments is on its way to create wonders. Malaria diagnosis on itself isn’t merely a simple method and the quick and correct detection of this deadly disease across worldwide is in fact a critical concern. We have gone through simple to create open-source methods leveraging Artificial Intelligence that will provide us progressive precision in diagnosing malaria thereby allowing Artificial Intelligence for medical good. Since AI is a rising machine learning procedure presently, we might hope several innovations to knock soon for functions like cell staging, cell classification and cell segmentation within automatize malaria detection.

Leave a Reply

Notify of