“We spoke to the health economists.”
Google’s AI product manager, Lily Peng, on how to implement an AI screening program in a under-resourced health system.
Imaging Wire Sponsors
- Carestream – Focused on delivering innovation that is life changing – for patients, customers, employees, communities and other stakeholders
- Focused Ultrasound Foundation – Accelerating the development and adoption of focused ultrasound
- Medmo – Helping underinsured Americans save on medical scans by connecting them to imaging providers with unfilled schedule time
- Nuance – AI and cloud-powered technology solutions to help radiologists stay focused, move quickly, and work smarter
- POCUS Systems – A new Point of Care Ultrasound startup, combining a team of POCUS veterans with next-generation genuine AI technology to disrupt the industry
- Qure.ai – Making healthcare more accessible by applying deep learning to radiology imaging
The Imaging Wire
Biomechanic Imaging
University of Auckland researchers are developing a biomechanical analysis technique that automatically merges information from different breast images to personalize and automate breast cancer diagnosis and treatment. Here are some details:
The Tech – The machine learning-based technique will provide clinicians with more information about abnormalities, as lesions and other signs of cancer can appear differently with different imaging technologies (mammo, MRI, US) or when patients are positioned differently (standing, sitting).
The Benefits –Clinicians might be able to improve diagnostic accuracy by using biomechanics to co-locate micro-calcifications (via mammogram images, generally captured while standing) and regions of increased blood supply (via MRI images, generally captured while laying down). The technique may also help surgeons locate suspicious lesions, improving the effectiveness of tumor extraction surgeries.
Next Steps – The technology is still in development, but it’s making progress towards becoming a “real-time clinical application.”
Biopsy Selection Breakthrough
MD Anderson researchers developed an image-guided breast biopsy patient selection protocol that can reliably predict residual disease in breast cancer patients and may be able to help identify “exceptional responders” who may not require surgery. Here are some details on a breakthrough that could “be a major step toward improving quality of life” for patients with breast cancer:
Avoidable Surgeries – Data from 166 women revealed that neoadjuvant chemotherapy (NAC) completely eradicated cancer in 51.2% of cases, suggesting that follow up surgery is unnecessary for many women. However, they also found that image-guided biopsy had a high 18.7% false negative rate (FNR) when measured across the entire group.
Selection Protocol – MD Anderson found that using vacuum-assisted image-guided biopsy on patients who are particularly likely to respond to NAC (n=76 women with <2cm abnormalities) resulted in a much lower 3.2% FNR, 97.4% negative predictive value, and 89.5% overall accuracy.
Up Next – The team is planning future studies focused on deescalating treatment for women identified as NAC-exceptional responders and evaluating the safety of eliminating surgery for these patients.
The Wire
- The UK NHS is headed towards a radiographer staffing shortage as a decline in applicants is causing some of the country’s universities to stop offering radiography degrees. The decline was attributed to the end of a scholarship that was effective at attracting older learners to radiography and other healthcare careers (also: mental health nursing, learning disability nursing, podiatry, and prosthetics). Many students now view the combination of £9,250 in annual fees and 37hr/week unpaid NHS work placements as unfair.
- A study in the Journal of Nuclear Medicine revealed that the 89Zr-Immuno-PET whole-body imaging technique can predict cancer treatment effectiveness by measuring monoclonal antibodies (mAbs) in normal tissues and tumors, rather than using invasive tissue or blood sampling methods. The study scanned 36 patients 1-7 days after they were injected with 89Zr, finding that it improved tumor detection (particularly on the first day) and could be used to predict which therapies will be effective and/or cause toxicity.
- Australian radiology AI startup, Harrison.ai, landed nearly $20 million in funding that it will use to grow its team (targets: data science, software engineering, clinical, and medtech backgrounds) and develop new AI applications for radiology, pathology, and hospital operations. Harrison.ai has been a local success story due to its physician-led structure/philosophy, bootstrapped operations (before now), the progress of its IVF effectiveness solution, and its new partnership with I-MED (the largest Aussie imaging center company) to create imaging AI solutions.
- A study presented by a Chinese team at RSNA found that although both automated breast ultrasound (ABUS) and handheld ultrasound are valuable additions to mammography for breast cancer detection, ABUS is significantly more effective. The study screened 1,266 women (40 – 69yrs, 323 BC cases) with mammography and the two ultrasound methods, finding that DM+ABUS and DM+handheld US had similar sensitivity (0.988 vs. 0.985), while DM+ABUS had higher specificity (0.876 vs. 0.857) and higher AUC (0.932 vs. 0.921).
- Senators Cory Booker and Ron Wyden urged the U.S. CMS, FTC, and major U.S. insurers to protect against racial bias in healthcare AI algorithms, pointing to a recent study that found racial bias in a population health management algorithm. The algorithm confused minorities’ lack of healthcare access/costs with greater health.
- A study on breast cancer malpractice lawsuits (n= 253 cases, 2005-2015) published in Clinical Imaging found that lawsuits were more likely to include younger plaintiffs (46 median age), most commonly involved radiologists (43%, followed by surgeons w/ 27% and OBGYN w/ 26%), often concluded with multimillion-dollar payouts ($978k - $2.3m), and were most commonly due to delayed diagnosis (82%).
- The state of Missouri recently held a joint task force hearing to evaluate whether it should require radiologic technologists to maintain a license (only 5 states don’t require RT licenses), with some arguing that licensing is a safety necessity and others saying that RTs can learn on-the-job and new licensing requirements would hinder recruitment. Missouri previously evaluated requiring RT licences in 2016, but was unable to finalize a law.
- A UPenn study confirmed that using a call triage assistant, who answers phone calls to the main reading room during the busiest hours (12-7pm weekends), improved workflow efficiency and reduced resident stress without changing error rates. The three-month study found that the majority of the 168 telephone calls could be handled by the call triage assistant, reducing resident interruptions by 71% and improving study turnaround times (44.3 min vs. 75.2 min), while rates of minor (7.5% vs. 6.7%) and major discrepancies (0.4% vs. 0.2%) were similar.
- Just when it appeared surprise medical billing legislation was going to negatively impact radiologists, the U.S. House Ways & Means Committee introduced a new framework that even the ACR is “encouraged” by. The new framework would ensure that patients treated by an out-of-network provider at an in-network facility only have to pay in-network rates, while respecting “the private market dynamics between insurance plans and providers” by allowing them to come to an agreement on their own before using a structured process to settle payments.
- An article from a UPenn and Seton Hall team detailed how informatics-based solutions can help alleviate radiologist burnout by reducing time spent on noninterpretive tasks, decreasing interruptions, facilitating connections with colleagues, and improve patient care. The solutions include adopting virtual rounds (interaction with ICU teams via web cams), linking the PACS and EHR, adopting a deconstructed PACS based on specific applications, using EHR pop-ups with relevant information, automating the dictation process with pre-populated fields, adopting AI, and leveraging radiomics.
- Siemens Healthineers announced the FDA clearance of its 64-slice SOMATOM go.Sim and 128-slice SOMATOM go.Open Pro RT planning CT systems, three months after debuting them at ASTRO. The systems update and expand Siemens’ single source CT “SOMATOM go” RT planning lineup (previously limited to SOMATOM go.Up for RT), adding a larger 85cm bore and new patient marking, patient setup, and organ contouring features, as well as the SOMATOM go.Open Pro CT system’s Direct Intelligent 4D tech (adapts image acquisition to a patient’s breathing to optimize image quality).
- New research (n=9,800 women, Trial 1: 1999-2002, Trial 2: 2007- 2011) finds that U.S. women who rely on Medicaid or Medicare for coverage have higher breast cancer mortality rates (Trial 1: 1/3rd higher, Trial 2: 50% higher), are less likely to participate in clinical trials, and are more likely to stop participating in clinical trials earlier (1/4th stopped early vs. 1/7th) compared to women with private coverage.
- A study presented at RSNA found that 18F-FDG PET/MRI is more effective for nodal staging breast cancer patients than MRI. The study scanned 56 women with newly diagnosed breast cancer (25 with lymph node metastases), finding that lymph node metastases was identified using axillary PET/MRI among 76% of the women, versus 56% with breast MRI, 60% with axillary MRI, and 72% with axillary sonography.
The Resource Wire
- The Focused Ultrasound Foundation recently awarded a University of Virginia team a grant for research on MRI-guided focused ultrasound as a new, incision-free, radiation-free treatment option for lung cancer.
- Qure.ai was highlighted in this Hardian Health blog for the significant research supporting its AI algorithms.
- This Carestream blog listed the top eight reasons healthcare providers should make the shift to digital.
- This Nuance video shares how Orlando’s Arnold Palmer Children’s Hospital used PowerShare to Ditch The Disk and improve patient care.
- Did you know that imaging patients are most likely to no-show for their procedures on Mondays and Saturdays? By partnering with Medmo, imaging centers can keep their schedules full, despite the inevitable Monday no-shows.
- POCUS Systems’ forthcoming ultrasounds will combine ease of use, durability, and reliability, allowing clinicians to focus on their patients.
