Artificial intelligence to combat river blindness

In sub-Saharan Africa, black fly mosquito engraving infects humans with the larvae of the roundworm Onchocerca volvulus – the trigger of river blindness, the so-called onhocerciasis. First, the larvae grow into sexually mature worms that prefer to live in nodes under the skin. The female, which can grow up to 60 centimeters, produces up to 1,000 offspring a day, the so-called microfilariae. They spread through the lymphatic channels in the skin to the eye. That's where the cornea is inflamed, whose destruction leads to blindness. Also, "baby earthworms" are picked up by new blacks in their blood meal and are still expanding.

The WHO recommends treating everyone in the affected areas with the usual Ivermectin, which kills offspring of earthworms without harming adult earthworms. Despite this therapy, new generations of microfilariae can emerge relatively quickly, which are responsible for the continued transmission of mosquitoes and for symptoms of visual disturbances and dermatitis. Therefore, treatment must be performed nationwide over many years in order to eradicate the disease with a viable one. "It's important to find drugs that directly kill adult earthworms," ​​says Prof. Dr. Honey. Achim Hörauf. The Institute of Medical Microbiology, Immunology and Parasitology (IMMIP) at Bonn University Hospital is under his leadership one of the leading institutions in the development of such new therapies.

Antibiotics in the fight against earthworms

The key could be the symbiosis discovered by Prof. Hörauf. The nematode Onchocerca volvulus has been home to the bacteria it needs to survive for millions of years. If these bacteria die, the parasite will die sooner or later. "Antibiotics with a spectrum of action that are specifically targeted to these bacteria are therefore an opportunity to permanently prevent the transmission of river blindness," says Prof. Hörauf. Antibiotic doxycycline, a prototype drug discovered by the Hoeruf group, has been used worldwide to treat filaria in clinics, but is not well suited for mass treatment in remote areas with poor infrastructure. Because the antibiotic must be taken daily for four to six weeks to develop its effect. As part of the worldwide research alliance, Bonn parasitologists continue their quest for faster action with shorter duration of therapy, eventually ending the nematode Onchocerca volvulus. The Bill & Melinda Gates Foundation has long funded this task.

The goal of AI is to accelerate the development of anti-earthworm therapy

However, to evaluate the success of such treatments, the nodes must be removed from the patient's skin and analyzed. Based on the obtained thin – so-called histological – sections, experts at the microscope evaluate the sex, vitality, embryonic development of roundworms and the presence of symbiotic bacteria. "This process is time-consuming and directly depends on the experience of the assessor," said Dr. Ute Klarmann-Schulz motivation, why her interdisciplinary working group at IMMIP was tasked with optimizing this analysis process. "With Dr. Daniel Kühlwein, a member of the Center for Excellence in Artificial Intelligence at Capgemini, a global consulting firm, he has been able to hire an expert in the field of artificial intelligence for collaboration."

With his support, a working group led by Klarmann-Schulz at Bonn University Hospital wants to develop an AI system that will automate the evaluation of histological sections. In doing so, they revert to already existing deep learning models for object recognition. "Pre-trained AI systems can basically already see, that is, they can recognize lines for example," explains Kühlwein. "Through transfer learning, we're empowering them for our new special application." The AI ​​system learns to identify and evaluate worms from already evaluated microscopic images of histological sections from numerous clinical studies that Bonn parasitologists have already conducted with their African partners. Scientists want to reduce the time it takes to evaluate and establish an objective standard for analysis.

Further information is available at:
https://gcgh.grandchallenges.org/grant/deep-learning-ai-histology-onchocerciasis
http://www.microbiology-bonn.de/immip/de/about
https://www.capgemini.com/de-de/service/kuenstliche-intelligenz-ki-ai/

Media Contact:

Dr. Ute Klarmann-Schulz / Prof. Dr. Med. Achim Hörauf
Institute of Medical Microbiology, Immunology and Parasitology
Bonn University Hospital
Phone: 0228 / 287-14612 (Klarmann) or 15673 (Hearing)
Email: Ute.Klarmann-Schulz@ukbonn.de; Achim.Hoerauf@ukbonn.de

Dr. Daniel cold wine
Center for Excellence, Insights and Data of Artificial Intelligence, Germany
Capgemini | Cologne
Mobile: 015140251432
Email: daniel.kuehlwein@capgemini.com

idw 2019/08