“Until one has loved an animal, a part of one’s soul remains unawakened.” –Anatole France.
Cancer is a kind of pandemic, which we have been witnessing for ages. It is a disease outbreak that we have already accepted because of its deadly outcomes.
Fact: “Nearly 20 million including humans and animals have died of cancer in the past two years. This is five times the number who have died to date of Covid-19.”
The painful journey of cancer patients and their surrounding ones are dreadful. Early detection is a critical feature in the case of the most complicated epidemic diseases like cancer. Therefore, early prognosis and selection of treatment protocols is the key. Some grassroots level problems actually need urgent attention (below mentioned in mind map). With the use of the latest technologies such as artificial intelligence, the healthcare industry could get deeper insights from the data.
Integrating empathy with AI models and the latest technologies can widen its fields of knowledge and provide further and more advanced solutions to complicated problems. Solving problems for all kinds of species, whether it’s humans or animals, can actually be counted as the contribution of technology to save lives.
We have witnessed the outcomes of some epidemic diseases such as cancer, but now it’s imperative to focus on it. Combining human cognitive skills and expertise with novel AI techniques can lead us to the most promising solutions for complicated health problems and are scalable to all species.
Need of AI in Cancer Prognosis
To explore the supportive role of AI technology to work on various aspects of cancer with the medical experts, a mind map has been designed to give an overall perception of the use of technology to save lives.
Mind map: The use of artificial intelligence in the field of oncology
Based on current scenarios from the above-mentioned mind map, the screening and diagnosis of cancer are the most crucial areas to work on. For developing an efficient screening system and early diagnosis of cancer, especially for pets — because they can’t tell about the symptoms or feelings verbally — observation or visual inspection plays a key role.
Generally, visual inspections can be categorised into symptoms; physical appearance; and behavioural change. Some of the acute symptoms which can give an early clue for medical examination and hence can be formulated to create an augmented system for diagnosis, screening and clinical decision-making systems are — lumps (big or small); swollen lymph nodes (neck, underarms, below thigh joints); dark colour tongue; recurrent blisters and wounds; heavy panting; extreme tiredness; rise in WBC count (laboratory variable); sudden high temperature; blood in the stool or unexpected bleeding; diarrhoea; and slow wound healing. In terms of physical appearance section, one can observe red or brown rashes on the skin; dry nose; and sad or panic facial expressions, and in Behavioral changes, one can keep a note on their pets being moody, irritating etc. (Disclaimer: can be seen for other comorbidities)
Example Problem Statement
An AI-based solution to detect malignant or non-malignant tissue and its type within an animal’s body?
Why is an AI-based system required?
Manual process: According to oncologists and pathologists, the first step in the cancer diagnosing process is to analyse the histopathology slides of the tissues, which is the most crucial and critical part of the diagnostic system. So, typically, after visualising some common symptoms, laboratory tests can be performed. According to medical experts, one of the manual methods of diagnosing cancer is to do FNAC at the first step and thorough histopathology slides analysis — i.e. the study of tissue through microscopic view to detect the disease under the supervision of a medical expert.
During this histopathology process, the medical practitioner uses certain patterns like horizontal & vertical zigzag to analyse tissues on slides. However, by choosing certain patterns on sample slides rather than the whole slide, some important information might be missing contributing to polarity results. Hence, results or outcomes are less precise. And that is where automation comes into the picture to eradicate human biases and errors. Consequently, this can help in precision medicine and can result in a better selection of treatment protocols. Perhaps, an augmented or artificial intelligent model can provide more accurate outcomes in less time.
How to develop an AI-based self-diagnosing system to detect malignant or non-malignant tissue and its type within an animal’s body?
Automation: For an AI-based solution, histopathology slides information is needed to be digitised first. Sample slide images should be converted into a digital version and recorded as an EHR (Electronic Health Record). This digital information (an image or pattern) from histopathology slides will be analysed using heuristics and biological parameters (collected from medical experts). The Digital images from EHR can be used as an input for the self-diagnosis system and subsequently for classifying malignant and non-malignant tissues. For critical cases, supportive information like sonography images and X-ray images can be considered.
A holistic approach for the development of an AI-based self-diagnosing system to detect malignant tissues and their type in an animal’s body has been proposed in the below-mentioned figure:
Using a Deep Learning(DL) model, an image corpus of benign/ malignant tumours is collected and given as an input to the model to detect genetic and molecular tumour cells and their alterations. Once an image of a patient’s histopathology slides images are analysed, its features are extracted and classified. Morphological analysis of tumour cells is observed and matched with the training corpus. Biological vitals stats are matched with the knowledge base of all kinds of lymphoma (type of cancer) to validate the outcomes for the additional or supportive information.
This is just an outline for a kind of AI-based solution to save the lives of animals from deadly diseases like cancer (lymphoma). There are many such ways and areas to use the latest technologies to better animal life with the right focus.
Research Perspective
Research conducted at the interface between AI and animal healthcare requires strong interactions between animal biological domains, i.e. infectiology, immunology, clinical sciences, the study of genomes, epidemiology, and veterinary sciences, and data science domains such as data analysis, statistics, precision medicine, drug discovery, predictive models and reasoning, together with highly efficient veterinary medical experts.
The need for research, training and support are crucial issues at national, European and international levels. Also, a facilitated and trusted connection is required between researchers, medical experts, and technology industry partners, who are often the holders or collectors of data of interest to solve Animal Healthcare (AH) research questions through AI approaches.
Wrapping Up
The amount of time and money invested into cancer drug research, development, and clinical trials have continually increased over the past few decades. Despite record-high cancer research and drug approval rates, cancer remains a leading cause of death. The use of the latest technologies in the Animal Healthcare(AH) domain is marginally low. Many animal lives can be saved by focusing on implementing life-saving technology solutions or programming empathy for animals in veterinary healthcare departments or hospitals.
With this article, I am trying to highlight the main areas that need urgent attention from technology perception to fight against a deadly disease like cancer for all kinds of species.
The development of AI skills within the AH community is limited in relation to the needs. Opportunities for collaborations with AI teams are limited because these teams are already in high demand. A training effort must be provided and generalised to ensure that AH researchers are well aware of AI’s opportunities and limitations and AI approaches’ limits and constraints. Finally, the current boom in AI now makes it possible to integrate the knowledge and points of view of the many players in the field of animal health and welfare further upstream.
However, this requires that AI and its actors accept to deal with the specificity and complexity of AH. The library of AH-related knowledge shouldn’t only be used for some insight gains for the betterment of the human species and use this information for designing and developing AI tools for animal healthcare and the betterment of animals.
*FNAC is a diagnostic first investigation which gives differentiation of malignant and benign nature of the cancer tissues.
