Latest developments in predictive analytics, and democratization of AI tools have led to a revolution in medical diagnosis. The benefits have trickled down to the large communities of academia, budding students and professionals who would have been otherwise excluded from contributing to this domain, and millions of enthusiasts like us, who might have remained ignorant of this amalgamation of these seemingly unrelated disciplines - computer science, statistics, and medicine.
Much less visible to the public eye are the clinical imaging companies, which have created their entire service models around the ongoing AI revolution. These organizations rely on inhouse resources to develop and validate the technology behind the models. But these organizations are still far from removing the complete reliance of medical specialists in detecting ailments, but are proving to be competent assistants. The FDA also has taken effective measures to regulate the AI based Software as a Medical Device (SaMD), especially to assess the risk to patients, ensuring transparency in the complex process of AI enabled process product development.
Arguably there is no place better than open data science competition platforms such as Kaggle, Drivendata etc. For example, Open Source Imaging Consortium (OSIC) launched a competition on Kaggle to test various Artificial Intelligence algorithms to detect Pulmonary Fibrosis Progression. The competition is still going on going, with more than 1000 teams participating. The ailment has no known cause or cure, which makes it for a very challenging prognosis. In order to find a solution, OSIC ensured that the problem statement reached to a larger audience with a diverse skillset. OSIC has provided the participants with huge datasets that comprise of patient data such as patient condition, lung capacity, smoking status etc.
Another interesting competition that took place recently on Drivendata is Clog Loss: Advanced Alzheimer’s Research, which is sponsored by a science community – Stall Catchers. The competition was about developing a machine learning model to detect clogged blood vessels in the brain, which could be the reason behind the Alzheimer’s disease. The 3D image stack is provided by on 3D image stacks provided by Cornell University’s Department of Biomedical Engineering, which was to be used to train and test the AI models.
At Choice TeleMed, we have the expertise to enable AI based medical imaging companies to develop or integrate existing highly performant technology-based tools and navigate effectively through the complexities of medical regulations by leveraging our wide network of specialist medical doctors. If you are thinking about integrating AI into your Imaging workflow or incorporating AI read into your clinical trial (for a variety of reasons) our team of experienced professionals would be delighted to have a chat.
Owing to a number of attractive traits, the Asian continent is well positioned to become the preferred destination for cost-effective clinical trials. This can be demonstrated by a 7-fold increase in the number of registered clinical trials in Asia over the last decade (Ali et al., 2019). First, Asia is becoming increasingly attractive as a clinical trial destination due to its presence of heterogenous population groups. Asia is the most populated continent in the world, with 4.4 billion people living on the continent alone – more than half of the world’s population (Statista, n.d.). Second, Asia has access to larger patient pools providing a comparable environment to carry out clinical trials.
For instance, pharmaceutical companies would be more likely to conduct liver or gastroesophageal cancer treatment trials in Korea or even in China, because these countries provide a large pool of patients for conducting trials for these diseases (Lee et al., 2014). Third, Asia offers lower costs for clinical trials, which are between 30 % and 40 % lower than in Western countries. Lastly, Asia has knowledgeable key opinion leaders (KOLs) and experts across a wide range of therapeutic areas. For example, China alone has 17 academic institutions and hospitals that have made a significant contribution to regenerative medicine research and has over 70 active trials.
At Choice we recognise the benefits UK Trials Imaging expertise can offer to China and East Asia. Our USPs are having innovative global imaging platform technology, and a vast network of leading UK Specialist Consultants with extensive experience in trials and research. An example of how Clinical trials in Asia could benefit is by completing one scan read in the UK. Benefits in this example would include an additional layer of expertise, mitigating bias and demonstrating independence through peer view from the UK, adding monetary value to trial outcomes, and subsequently enhancing your organisational value. Hence, Choice TeleMed are an ideal UK partner to complement your clinical trials in Asia.
If you are attending ChinaBio August 25-27th, we'd be delighted to (virtually) meet up and discuss further.
Over the past 20 years the pace of technological developments in medical imaging has been relentless, improving our understanding of disease processes and
better quantifying treatment effects. Advanced imaging techniques in computed tomography (CT), magnetic resonance imaging (MRI) along with functional imaging technologies such as positron emission tomography (PET) hybridised with CT or MRI acquisitions, coupled with molecular tracers to assess metabolic processes has led to annual increases of over 10% in Radiology output (with similar increases in complexity) and has ultimately delivered radiology sub specialisation.
The UK healthcare system has been at the forefront of these patient care innovations, exemplified by the central role imaging departments now have in the patient journey by enhancing diagnostic accuracy, assessing treatment responses and the role of imaging based screening for disease; such as high risk smokers by CT scanning for early asymptomatic lung cancer. The UK radiologist sub specialist workforce has attained a position as a credible and internationally respected community shored up by high standards of training, continuing professional development and the governance/regulatory framework provided and enforced by the Royal College of Radiologists and the General Medical Council.
Not only has medical imaging and the field of radiology benefited from the contributions of countless physicists, mathematicians, material scientists and molecular biologists to name but a few, IT and the Internet has enabled an explosion in Teleradiology, attracting radiologists with remote working, pay per fee models and ability to sub specialise for more than one source. For CROs, this has given greater access to uploading trial imaging, but perhaps more crucially, access to leading trial reporting specialists renowned for publications and research.
Subsequently, when a clinical trial demands the highest in quality & credibility, where better to have the radiology element reviewed than in the UK.
The life sciences industry has recognised the valuable contribution medical imaging techniques can have in pharmaceutical or medical device clinical trials, imaging in clinical trials has grown by over 700% since the early 2000’s. With the growth of medical imaging in both pre-clinical and clinical phase research, imaging endpoints are now routinely requested by regulatory bodies, thus adding news layers of complexity to already busy trial schedules.
It is not surprising why there has been this astonishing growth of imaging in drug development, when one considers the non-invasive aspects, allowing tissues to be studied that are difficult to biopsy. The facilitation of precision medicine by virtue of screening and stratifying patients into those groups who are more likely respond to the therapeutic product under test, usually in the paradigm of molecular level disease setting. Hence drug development timelines can be shortened, overall budgets are reduced and regulatory approval pathways accelerated.
The imaging aspects of a clinical trial will involve the scanning of patient volunteers to strict protocol guidelines, these are often archived at the local scanning centre/hospital institution as well as being exported to a single site known as the core imaging laboratory for a centralised review. There is full anonymisation of the data sets with only code identification, these imaging data sets undergo reading by carefully selected experienced radiologists against the study criteria, often referred to as the imaging manual. Any measurements should be recorded, allowing an audit trail and the overall assessment of response is made against hard objective specifics laid out in the imaging manual. To maintain quality, many core laboratories operate a 2 + 1 model, where 2 radiologists will independently report on the same study, for a further radiologist to perform an adjudicator role by only submitting the analysis closest to perceived ground truth.
Imaging plays a role in every stage of the process of drug development. Currently oncology clinical trials are the largest utiliser of imaging, in phase I clinical trials initial evidence for anti-neoplastic effect can be sought by size reduction in solid tumours using CT scans. Also pharmacokinetic and pharmacodynamic data can be elucidated. In phase II studies, drug safety and efficacy within a defined population is augmented by imaging often using standardised response criteria. Once such is the response evaluation criteria in solid tumours (RECIST), allowing response outcomes to be categorised as progressive disease, stable, partial and complete response. RECIST is undertaken by summating dimensional measurements of lesions into target and non-target lesions and following these over serial scans on each patient. Hence it acts as surrogate endpoint in a clinical trial rather than the ultimate proof of anti-cancer drug effect of improved clinical symptoms and survival.
Tumour size measured before and after therapy is considered an accepted means of assessing treatment response. However it is known that many novel therapies do not necessarily result in immediate tumour size reduction, thus allowing the functional imaging modalities of PET-CT and PET-MRI to look for metabolic changes such as reduction of glucose metabolism in atum our site following the rapy prior to any anatomical change. Other biological correlates of tumour behaviour can be quantified via imaging namely blood flow, vascular permeability, cellularity and hypoxia utilising these functional techniques.
To maintain quality, in house proprietary software is often used by core laboratories to significantly improve the evaluation of clinical imaging trial data. For example image analysis software is used to guide a radiologist through the study read, pre-processing and segmenting regions of interest in keeping with the imaging manual, thus reducing deviations and reducing reader bias from creeping into the final analysis.
The remarkable progression in the field of clinical radiology and medical imaging has provided an invaluable contribution to clinical research. Through technology and systems advancement, the high quality and specialism associated with UK expertise has become much more accessible to international CRO’s in order to meet their needs and requirements in fulfilling the rigorous aspects of imaging in clinical trials. The life sciences industry has recognised the valuable contribution medical imaging techniques can have in pharmaceutical or medical device clinical trials.
"The life sciences industry has recognised the valuable contribution medical imaging techniques can have in pharmaceutical or medical device clinical
Artificial intelligence has the potential to influenced many aspects of the clinical trial process and design. Right from selection, recruitment and patient monitoring to analysing real world data and scientific information in reaching predictive outcomes, AI has made significant breakthroughs in clinical trial operations.
Clinical trials processes have significant scope for improvement particularly through the application of AI as the space is prone to many challenges, some of which are flawed study design, not much success to demonstrate safety, participant drop-outs and unsuccessful recruitment. Let us take a look at some of the technologies that make use of AI and machine learning within the clinical trial process and their impacts on the future of the same.
The field of medicine has witnessed quite a few advancements in the past few years especially with the emergence of artificial intelligence. Medical imaging has been hugely impacted in terms of efficiency and accuracy so much so that initial scaremongering predicted that radiologists will be replaced by artificial intelligence (when the truth is AI will be invaluably complimentary to Radiologists reporting). Rest assured, artificial intelligence is set to change the scenario of advanced medical services much more in the coming years. Increased uptake, more algorithms and big data access will play a prominent role in sourcing and transforming medical imaging, including clinical trials imaging.
Impact of AI in diagnostic radiology
DL and ML algorithms will be used to train the systems of radiology which will result in a state-of-the art infrastructure. Of course the training material involves incorporation of intrinsic aspects. Computer aided detection (CAD) in mammography is now combined with DL Convolutional Neural Network (CNN) algorithm to detect breast cancer lesions and this is progressive step in itself.
All these and much more are awaited in the field of medical imaging in the coming years.
Globally, Imaging is under severe capacity strain with an 11% (UK) annual increase in scan volumes (with added specialism and complexity) and reduced time to diagnosis targets -with no significant increase in the number of reporters. Hence the potential of Imaging AI to revolutionise and become the saviour of diagnostics as we know it. However, there is a lot of ‘talk’ surrounding AI, particularly in Imaging and Radiology. And this talk ain’t cheap!
Given the fact that Imaging AI is a relatively new and evolving innovation, some 50-100 ‘Imaging AI companies’ have raised 100s of millions (if not billions by now) in VC, despite there being only 7 FDA and CE approved algorithms at the beginning of 2019. That figure has at least doubled during 2019, which demonstrates the speed of current advancement.
Our teleradiology modality coverage includes plain film X-rays, CT, MRI, and Nuclear Medicine, such as SPECT-CT, PET-CT, and PET-MR. Our sub-specialists include leading academics and award winners with multiple publications having worked at some of the best sites in the UK.