For decades radiologists have struggled with their workflow when using PACS to report studies. Whether it be the wrong studies showing up in their worklists or displays not showing the series in the way they prefer, radiologists have had to manually intervene to make headway against the ever-increasing number of studies being generated. These productivity issues are exacerbated when PACS administrators are then called in to fix the issues raised, taking them away from higher-value tasks to deal with mundane tasks. These should be issues that were resolved years ago but they stem from a basic problem: a lack of data governance. Data governance is no easy task to retroactively address and is only compounded with the advent of AI algorithms making workflow even more complex.

The Scandinavian leaders of AI in breast imaging presented their research at the ScreenPoint symposium at EUSOBI 2022 in Malmo, Sweden. Dr Kristina Lang presented the MASAI trial, the first prospective randomized controlled trial on the use of AI in breast screening as an alternative for double reading. Based on her previous retrospective studies, she is convinced that AI could lead to a more efficient and more effective screening programme. In the MASAI trial at Unilabs/Skane University Hospital Malmo, women are randomly assigned to a control arm where exams are double read as usual, or to the AI-based intervention arm: Transpara triages screening exams based on risk for malignancy and assigns 90% of all screening cases to single reading, and 10% to double reading. In addition, the top 1% most suspicious cases are automatically recalled.

AXREM is the UK trade association representing the interests of suppliers of diagnostic medical imaging, radiotherapy, healthcare IT and care equipment in the UK. The group has formed the AXREM AI SFG (special focus group) to promote the adoption of beneficial AI technology and to identify and seek to overcome hurdles to adoption encountered by suppliers. Most recent AI solutions are based on a process of deep learning in which an artificial neural network is presented with annotated examples of a particular kind of pathology and gradually learns to recognise what that pathology looks like. This process is data dependent, and so the work of data scientists and curators is as important as the AI scientists and software engineers who develop the actual algorithms. A key part of developing safe and reliable AI solutions is the use of a representative dataset while performing the AI training.

The breast imaging service at Leeds Teaching Hospitals NHS Trust is the first in the UK to introduce an AI technology from Densitas that evaluates the technical quality of mammograms. The Canadian software company has worked to customise the intelliMammo software to meet the department's needs and the image quality standards required by the NHS Breast Screening Programme. The technology will be implemented in the screening and symptomatic breast services to provide mammographers with instant feedback and AI-generated positioning information. This means that any quality issues are flagged at the patient's appointment, which should reduce recall rates and enhance the accuracy of the image interpretation. Director of breast screening Dr Nisha Sharma said: "Using AI will make a real difference to the quality of our breast screening programme and ultimately to the health outcomes of women in Leeds. We are delighted to collaborate with Densitas and build on this foundation for future developments that will improve breast care."

Zegami provides an image-based data visualisation platform designed to enable users to explore large image datasets in order to unlock insights and build machine learning models. The company points out that, with any system that is reliant on data, overall effectiveness is dependent on the quality of data that it utilises. If the data is good, the value of output will reflect this, and AI is no different. In machine learning-based models, when trained on incorrect, underrepresented or biased data, the models can Themselves become biased. "In the field of AI, we typically encounter five different types of bias: algorithmic, sample, prejudice, measurement and exclusion bias. These can be difficult to eliminate, particularly as certain biases may be unconscious," states Zegami.

The Golden Jubilee National Hospital in Clydebank, Scotland, has updated its cardiac CT scanning equipment. The Aquilion One/Prism Edition CT scanner introduces next-generation reconstruction technology to assist with high cardiac imaging case loads. "We are delighted with the new CT scanner," said CT superintendent radiographer Julie Morrison. "It is easy to operate and our radiologists have been delighted with the fantastic quality of cardiac images at low dose to support our heart and lung patient services. The wide area detector will enable us to obtain an entire heart in a fraction of a second and the AI technology further enhances the ability to gain low dose, high quality images at speed. We feel ready for the future."

As 24/7 imaging activity rises in volume and complexity and the workforce deficit grows teleradiology management of the acutely unwell patient has become increasingly important. The concept is not new: GMC registered reporters with home workstations, potentially anywhere in the world, interacting seamlessly with client RIS systems. Faster data transfer, strict information governance protocols and evidence-based imaging have enabled developments of this service. As the UKs largest teleradiology company, Medica can offer bespoke tools and pathways to enhance patient care. For major trauma cases reporters provide a'seriously injured patient' report within 10 minutes of image receipt direct to the trauma team leader, enabling prompt management of life-threatening injuries.

A chest x-ray AI decision support tool for radiologists and clinicians that detects 124 clinical findings has been launched. The study found that when used as an assist device, Annalise CXR significantly improved the ability of radiologists to perceive 102 chest x-ray findings in a non-clinical environment, was statistically non-inferior for 19 findings and no findings showed a decrease in accuracy. It also assessed the stand-alone performance of the model in a non-clinical environment against radiologists in identifying chest x-ray pathology, as well as investigating the effect of model output on radiologist performance when used as an assist device. Annalise CXR's AI model classification alone was said to be significantly more accurate than unassisted radiologists for 117 of 124 clinical findings predicted by the model and was non-inferior to unassisted radiologists for all other clinical findings. Co-founder and ceo Dimitry Tran said: "A major challenge facing global health systems is that the number of scans requiring clinical interpretation is growing at a greater pace than increases in the number of radiologists to interpret them.

Developer of AI systems for breast care ScreenPoint Medical BV announced the introduction of Transpara powered by FusionAI, an improved and enhanced version of Transpara 1.6, at the European Congress of Radiology virtual meeting in March. Transpara has been in use in more than 20 countries and ScreenPoint says that the latest clinical studies reveal major benefits for radiologists and patients: "Up to 35 per cent of examinations with interval cancers found on earlier mammograms and, to help reduce workload for increasingly pressured radiologists, up to 70 per cent of examinations can now be confidently labelled as normal. Overall Transpara has been shown to match the performance of specialist breast radiologists in both 2D and 3D studies." Transpara with FusionAI is up to 28 per cent more accurate than previous versions, helping to detect more cancers and earlier, the company added. See the full report on page 26 of the April 2021 issue of RAD Magazine.