This article is based on the 2017 research conducted by Alcantara et al (2017). The research was primarily focused on developing nations, in particular, all the work was based on Peru, a country in South America. Peru is not a rich country, and hence the availability of advanced gadgets for detection, analysis, and diagnosis is not that trivial in Peru and such countries.
The key issue in the paper [1] was to mitigate the TB epidemic. It is known that technological development heals the diseases. One of the differences between poor nations, developing nations, and developing nations is the fact that in developing nations it does not take as much time to claim a TB from an X-Ray. Not only are AI-based software in developed nations smart, but they are also accurate and reliable. It is noted if this disease is administered and detected in a timely way, then the recovery can be smooth and proof-free.
A detailed analysis has been presented in the paper [1] about how many X-rays of patients were initially present, to be used in AI software, and how many frames were added to X-rays to detect the region of the lungs that is affected by the disease of TB. Then the methodological process of building more X-Rays, and especially annotated images which can be information. Such images can be diagnosed in mobile devices, to start with for ease and continuity of better healthcare. The approach is storage, training, annotation, and then finally review.
Once annotated images were obtained deep learning-based algorithms were proposed to be applied on newer data, called the testing data, by keeping 4/5th of the images for training. The final dataset has 4701 X-Ray images out of which 453 healthy normal X-Ray images and 4248 T.B. infected X-Ray images. GoogleNet model from Caffe were used in two kinds of experiments (multiclass and binary classifiers)
However, time has passed since 2017 and the current state of the art in countries like Peru is still required. What drew my attention to Peru was a question asked me about South America, then, which intrigued me to read this paper in depth.
Future Works from My Desk (Me)
Apart from collecting X-Ray images, from places in Peru, one should consider for international health standards organizations (such as WHO) and standards. Since T.B. is not an isolated disease hence a need for global consolidated information, processing and diagnosis, Means the WHO or like organizations taking up guiding the research work, once done in say Peru, can be distributed in some other places, to be appended or accumulated, as the case may be, not just for T.B. but for other diseases as well,
Given vaccines and other medicines have been on the rise for T.B. since then, one thing is why not global eradications,
I think processes are much more synchronized, which means Peru did all this research, this research can be shared by others, unless the annotations and frames of the X-ray images change,
So aim is not about one country ahead of others, no aim is sharing all the global data or buying the licenses to access the annotation, and in most cases, these may be provided free of charge, as everyone wants to fight T.B. away.
Then, comes, steps to healing, medicines, and guidance, which should be paved free of charge, especially for poor or undeveloped nations,
Healing as a process, can be understood as an RNN (Recurrent Neural Networks) fact,
References
[1] Alcantara, M. F., Cao, Y., Liu, C., Liu, B., Brunette, M., Zhang, N., … & Curioso, W. H. (2017). Improving tuberculosis diagnostics using deep learning and mobile health technologies among resource-poor communities in Perú. Smart Health, 1, 66–76.
Artificial Intelligence, Computer Science & Mathematics 