2 4 R A D I O LO G Y I N E U R O P E Challenges in brain tumour segmentation under scope ‘Molecular diagnosis is better suit- ed, but is still a new and emerging field,’ she pointed out. Neuroradiologist Dr Sofie Van Cauter described the challenges to brain tumour image segmentation during the European Society of Medical Imaging Informatics (EuSoMII) annual meeting in Valencia. She also outlined how, when clinically validated, AI could help tackle such problems. Report: Mélisande Rouger The WHO classification of brain tumours has come a long way since first introduced in 1979. The 2016 classification was something quite revolutionary, neuroradiologist Dr Sofie Van Cauter explained during the EuSoMII annual meeting. ‘For the first time, molecular diagnostics were taken into account in the diag- nosis of brain tumours, leading to more detailed diagnoses and better reflecting the aggressiveness of the lesion and the patient’s prognosis and response to therapy,’ she said. ‘This has been made possible with the evolution of next generation sequencing.’ In her talk, Van Cauter focused on primary brain tumours and espe- cially gliomas, which are the most common type of brain tumours and come in different histological and molecular subtypes. In a first step, the pathologist determines the histological subtype and a grade of aggressiveness – the classic and still quite subjective, way to diagnosis. In the second step, as proposed by the WHO classification, several genetic mutations can be deter- mined, leading to a more detailed final diagnosis. Molecular diagnostics ‘We know, since we applied molecu- lar diagnostics, that we are much more able to predict the patient’s prognosis and response to thera- py. For example, in certain lesions a more benign histological sub- type can initially be considered. However, if the lesion has a certain gene mutation, we know the lesion will have an aggressive course and the patient’s prognosis is not good,’ Van Cauter explained. Typical segmentation of a tumoural lesion in the left frontal region as proposed by the BraTS consortium. Left Panel: A-B-C: whole tumour segmentation (yellow), segmentation of the tumour core (orange), the enhancing tumour portions (blue) and the necrotic core (green). Right panel: the different labels put on one anatomical image. Adapted from Menze et al. IEEE 2015. When dealing with a brain tumour, radiologists look at multiple MRI sequences performed in multi- modal imaging protocol. In recent years, neuro-oncology research has focused on advanced MRI tech- niques, which try to relate his- tological features in radiological phenotypes, such as cellularity or vascularity. In current clinical prac- tice, the basic imaging protocol consists of T1-weighted sequences, in which anatomy can be dem- onstrated, T2-weighted sequences, which relate to oedema and can be a measure of cellularity. There are also FLAIR sequences, to distin- guish more subtle oedema, and T1 contrast-enhanced sequences, which distinguish areas with disrupted blood-brain barrier from regions of non-disrupted blood-brain barrier. Diffusion and perfusion-weighted sequences provide additional meas- ures, reflecting vascular prolifera- tion and cell density. Challenges for brain tumour diagnosis and follow-up When a glioma is suspected, and when dealing with a brain lesion on MRI, radiologists still try to make a diagnosis according to the histologi- cal subtype. However, this diagnosis is quite subjective, which is a major issue. ‘Accurate, quantitative thresh- olds are not determined yet, we just base the diagnosis on qualitative, subjective features like “low-to-mod- erate oedema is probably reflecting this or that type of tumour”. We still have a problem with the diagnostic accuracy to predict the histological subtype of a glioma – it’s currently moderate to low. We hardly got started with the correlation between molecular profiling and imaging,’ she pointed out. Glioma therapy is a combination of surgery, chemo- and radiother- apy. There have been many efforts in more advanced therapy, but until now clinicians have stayed with the cornerstone of therapies established since 2005. An even more important issue in glioma is follow-up, currently done according to the RANO criteria, which try to classify patients in a certain category, from complete response to partial response to sta- ble disease in progression. ‘This is done, again, in a very basic way. Just to classify into four categories, you need to sum the perpendicular diameters of all enhancing lesions and compare them to the smallest measurable tumour volumes. The classification only takes into account T1-weighted sequences with con- trast administration and to a minor extent, features on T2 and FLAIR images,’ Van Cauter explained. Thus, radiologists face many chal- lenges in adequate diagnosis and follow-up of gliomas. Evaluation of gliomas is subjective; does not take into account robust quantitative parameters and is not adapted to the emerging evolutions in molecu- lar diagnostics. Alzheimer’s research: A lost century patients with dementia,’ he stressed. Medical imaging is needed for early and specific identification of patients at risk of a specific type of dementia before the onset of clinical Can imaging progress AD therapy studies? Lack of understanding around Alzheimer’s disease (AD) has significantly slowed advances in the treat- ment of this incurable condition. Imaging has proved to be reliable in differentiating between AD and other forms of dementia, and its contribution will continue to help develop profiling, an increasingly interesting approach for the development of new and more efficient drugs, according to Sven Haller, a Swiss neuroradiologist, who spoke during EuSoMIIs annual meeting, held last October in Valencia, Spain. Report: Mélisande Rouger Over a century of research in AD treatment has failed to cure peo- ple. For this, there are two main reasons, according to Haller, who will chair the New Horizons session on ‘Alzheimer’s disease and neuro- degeneration: visualising the invis- ible’ during this year’s European Congress of Radiology (ECR). ‘Giving therapy to only patients with clinical symptoms of dementia has greatly impacted on therapeutic advances. For 80 years, we’ve given drugs to patients who already had symptoms. However, we know that if we wait for symptoms to develop, by then 50% of the neurons are lost. This was already confirmed in a study from 1968 (Tomlinson BE, Blessed G, Roth M (1968) Observations on the brains of non- demented old people. J Neurol Sci: 7:331-356.1).Even if the medication works, it will very likely not revital- ise dead neurons,’ he explained. Another major setback has been the profound misunderstanding of Alzheimer’s disease, which is a form of dementia. There are many types of dementia, for example vascu- lar dementia, a group of frontal dementias and Lewy Body demen- tia, among others. Alzheimer’s and dementia are not the same, but they are commonly perceived as such, Haller pointed out. ‘You don’t have one stereotypical form of dementia. Not everybody who has dementia has AD. If you simply treat a per- son with signs of cognitive decline with AD medication, it may just not work, as this patient may not have Alzheimer’s disease,’ he said. Imaging central to early diagnosis and differentiating To complicate matters further, there are also different disease subgroups of AD, which must be tackled inde- pendently. Drug development must take this diversity into account, because there will not be ‘a magic universal drug that will work in all Multiparametric MR protocol allowing for the assessment of brain structure and atrophy, various markers of neurovascular diseases and functional brain perfusion in one single imaging session AI will help, but needs clinical validation AI entered neuro-oncology a few years ago. A Chinese group nota- bly developed an AI algorithm that determined the subtype of brain tumours much faster and more accu- rately compared to a group of neu- rologists. In radiology in general, the tasks in AI algorithms fall into three categories: detection, segmentation and classification. Segmentation algorithms consist of multiple steps. Lesions, or other tissue of interest, have to be diagnosed, delineated and finally, different tissue types must be determined. In the case of a brain tumour, solid tumour, necrotic core and oedema have to be separated from healthy tissue. The development of automatic seg- mentation algorithms that focus on these tasks is an enlarging field and in the last ten years, the number of publications has risen steeply. For segmentation, convolutional neural networks are used, which consist of several steps and multi- ple layers in a computer algorithm, using features related to symmetry, intensity gradients, etc. These char- acteristics are combined to clas- sify different regions as enhanc- ing tumour, non-enhancing tumour, necrosis or oedema. Until now, this work has stayed mainly in the field of engineering, signal and image processing. Using this kind of AI on a daily basis still requires multiple validation steps, from technical to biological and finally clinical validation. ‘Today, even starting from the beginning of the validation steps, we still have some limitations,’ she said. The first, very important techni- cal limitation is scarcity of data. The first general public studies on AI used data sets of 60,000 images or more, whereas in the medical field we have to work with smaller data sets – mostly between 100 and 1,000. Then, brain tumours are very heterogeneous – each is different symptoms. Imaging is a great way of assess- ing and differentiating dementia, which can further help to develop adequate medication for each dis- EUROPEAN HOSPITAL Vol 29 Issue 1/20