R A D I O LO G Y 1 3 Despite its power, validation remains challenging Relating AI to biomarkers Using artificial intelligence (AI) to push development of imaging biomarkers shows great promise to improve disease understand- ing. This alliance could be a game changer in healthcare but, to advance research, clinical vali- dation and variability of results must be factored in, a prominent Spanish radiologist advises. Report: Mélisande Rouger In clinical practice efforts are already ongoing to apply AI to obtain new imaging data and improve the step- wise development of radiomics and validation of biomarkers, Professor Luis Martí-Bonmatí, Head of Medical Imaging at La Fe Polytechnics & University Hospital in Valencia, told delegates at the ESR AI Premium event last April in Barcelona. ‘There’s a mass of imaging data obtained daily and waiting to be deciphered and ana- lysed to understand what’s happening in every patient across the huge diver- sity in people, diseases and disorders,’ he pointed out. Using genomics, proteomics and metabolomics, scientists can already unveil biological processes in an indi- vidual patient. Computational medical imaging enables evaluation of tis- sue properties and behaviours from medical images, to accurately describe things relevant to a patient at a spe- cific time. Using computerised model- ling, mainly thorough deep learning techniques, high-dimensional data can be extracted and mined to build descriptive and predictive diagnostic tools. Highly performing computation- al techniques, such as Convolutional Neural Networks (CNN), help to provide new information on tissue expression diversity from ‘real world’ imaging studies, making a noticeable contribution to advancing healthcare. ‘If we can implement computer- based processes designed to analyse medical images to depict and classify those tissue changes, with their value and distribution, we might have a nice tool in our hands to improve person- alised medicine using medical images,’ Martí-Bonmatí said. Not every radiologist is familiar with biomarkers, yet they could transform radiology clinical routine and deeply impact on healthcare. Biomarkers are similar measures to those obtained from blood samples; they may indi- cate biological processes, pathological changes or pharmaceutical responses. When biomarkers are imaged, sub- rogated features and parameters can be obtained that will give quantitative information on regional distribution of these changes whenever necessary. In other words, tissue changes can be depicted over time and located, mean- ing they are resolved in both space and time. As images do not harm the organs and lesions, researchers can evaluate heterogeneous distribution of whatever they want to look at, when- ever they want to look at it. Biomarkers can be used to diagnose phenotyping, so to detect or confirm the presence of a disease, or to iden- tify different diseases sub-types and decision to predict which therapy will work best,’ he suggested. Bonmatí said: ‘And then we will have biomarkers.’ In daily practice, huge amounts of images are generated from different modalities and sequences, using a broad diversity of information chan- nels for acquisition. Image prepara- tion, including registration, analysis, resizing, intensity normalisation and tissue segmentation, are the next step. ‘We need to virtually ‘take out’ the organ or tumour we want to evaluate, to check what happens in that tissue as automatically as possible. Once we obtain this volume of interest,’ he said, ‘we can move to picture feature attributes, which is mainly radiomics, and go into morphology or semantics, spatial distribution of signal intensity, and histogram distribution, with all Obtaining biomarkers is not an easy task and external validation in the clinical setting is key. ‘The relevance of those imaging findings and their correlation with the clinical endpoints and the impact on healthcare path- ways must be shown,’ he pointed out. For example, in texture analysis of liver metastasis, some parameters might enable classification of lesions into those that will respond and those that won’t, right from the onset of treatment. ‘That’s quite good for us.’ But the big challenge remains future variability – unavoidable because image acquisition parameters still dif- fer from one examination to another. ‘If we slightly change repetition time, even different habitats within a single lesion. They also can be used to assess why a tumour is responding to treat- ment, whilst other identical tumours are not, and to measure susceptibility to potentially develop a disease. When linked to treatment effect, biomarkers can have predictive value not only on therapy effectiveness, but also safety, by looking at the extent of toxicity, a well-known adverse effect. ‘Ideally, we could also have biomark- ers on prognosis and help determine likelihood of disease recurrence or progression, or patient survival, by taking a look at the lesions and organs where the abnormalities are present right at the beginning of treatment the different texture features that can be obtained across organs and tissues.’ Dynamic model parameters can also be used in the region or volume of interest, by calculating values that might describe what is happening in a given lesion through histograms, distribution statistics or spatial distri- bution of feature metrics. A huge amount of data has been generated at this stage, so efforts must focus on reducing data, and then applying statistics or multivariate analysis or classifiers like clustering signatures. ‘We might be lucky and link whatever we have here with the diagnosis, predictive or prognostic endpoints that are our interests,’ Martí- echo time, flip angle, slice thickness on an MR sequence, the obtained parameters will change. Processing techniques, methods, filters, quantifi- cation levels and image depth will also change the radiomic features,’ he said. ‘So, small changes in so many vari- ables will change the results.’ Signal dynamic parameters in MR techniques, such as intravoxel inco- herent motion on diffusion-weight- ed sequences, may offer additional parameters that can be linked to aggressiveness of prostate cancer, for instance. But, the variability with this approach needs consideration – num- ber, distribution and magnitude of the b-values, signal strength, amount of Parasites & company – what can the radiologist see? When pathogens travel with us Sunburn and happy memories are not the only things we can bring home from a holiday. Sometimes parasites, fungi, viruses or bacte- ria from distant countries accompany our return, later to become noticeable in unpleasant ways, often to pose a real health threat. At the German Radiology Congress in Leipzig, Dr André Lollert and colleagues ventured into the world of tropical and travel medicine. The senior paediatric radiology consultant at the University Medical Centre of the Johannes Gutenberg University, in Mainz, provided an overview on which of these ‘stowaways’ can also enter German hospi- tals – and how radiology deals with them. Report: Wolfgang Behrends In his Leipzig presentation, Dr André Lollert spoke about worms (hel- minths), such as Echinococcus and Schistosoma; fungi of the genus Histoplasma, and about viral and bacterial infections. These patho- gens mostly enter the body by ingestion (less frequently via skin injuries) and cause a wide range of diseases. ‘It’s mainly increasing tour- ism, but also migration that bring into our hospitals diseases whose origins lie in very different parts of the world. Therefore, the probabil- ity of being confronted with these exotic diseases in clinical routine is rising.’ Whilst native parasites such as fox or dog tapeworms are relatively common, the occurrence of melioidosis, a dangerous bacte- rial infection which is mainly found in Southeast Asia, is also on the increase. If patients come to hospital after being abroad, these cases often can be quite harmless – the clas- sic being travellers’ diarrhoea. ‘However, depending on the coun- try the patient visited, a more exotic, differential diagnosis must be con- sidered.’ For people whose immune defence is impaired by age or illness these pathogens can be life threat- ening. ‘Worms, in particular, move through the body once they have entered it, so different organs can be affected,’ Lollert explained. An image alone is not enough For the radiologist, the pathogens frequently remain invisible – only the traces they leave in the body can be seen. ‘Each parasite has a preferred part of the body where they settle. Whilst Echinococci often spread in the liver, histoplasmosis or melioidosis usually affect the lungs.’ Since the diseases may manifest in a very typical way, the interac- tion between radiology and patient anamnesis is vital, says Lollert: ‘Imaging alone does not deliver Continued on page 14 www.healthcare-in-europe.com Luis Martí-Bonmatí MD PhD serves as Head of the Clinical Area of Medical Imaging at La Fe Polytechnics & University Hospital, Valencia, Spain, and a member of the Spanish National Royal Academy of Medicine. He is also the founder of QUIBIM S.L. and serves as its Director of Scientific Advisory Board. noise and lesion type will all impact on the calculated parameters. Using voxel-enhanced dynamics, again in prostate cancer, the cel- lularity metrics obtained from the intravoxel incoherent motion, and the permeability Ktrans obtained through a pharmaco-kinetic model after contrast administration, may change widely in a clustered way. ‘If we perform a multivariate, multipa- rameter map, we can visualise those areas with high cellularity and high vascular permeability, which are the most aggressive ones. Unfortunately, the variability using these multivari- ate techniques together increases exponentially,’ he explained. Therefore, researchers should ask themselves a number of questions: do they have the right and precise answers by using texture analysis or feature properties or parameters from signal dynamics? Will this process prove useful in clinical practice? Why are the best answers to the right ques- tions still wrong? ‘Wrong means that we have a huge amount of variability in what we are doing. So, we have to look for the uncertainty of our truth. And we have to recognise that for anything we measure from images to be repre- sentative of a physical reality, we must have a clear relationship with the real- ity we are measuring. Imaging signal comes from voxels, and voxels have a huge amount of complex inner struc- tures, all with different properties and components, obtained with different protocols, techniques, and machines. So it’s close to impossible to have a standardised image processing or image acquisition or parameters. Once we recognise this,’ he concluded, ‘we can work on ways to improve our work.’ Dr André Lollert is a senior consultant in paediatric radiology as well as radiation safety officer at the University Medical Centre of the Johannes Gutenberg University in Mainz. Whilst his research focus is on paediatric radiology he also specialises in the diagnosis of metabolic and infectious diseases. In 2017, Lollert was awarded the Publication Prize of the Society of Paediatric Radiology for his cancer research work in the field of quantitative imaging procedures.