|
Complementary PE AI | Future Clinicians
March 23, 2022
|
|
|
|
|
Together with
|
|
|
|
“I try to read every Imaging Wire, and if I don’t have time, I at least make sure to check out the quote at the top.”
|
|
High praise from a loyal TIW quote reader at last week’s HIMSS show.
|
|
|
|
A new European Radiology study out of France highlighted how Aidoc’s pulmonary embolism AI solution can serve as a valuable emergency radiology safety net, catching PE cases that otherwise might have been missed and increasing radiologists’ confidence.
Even if that’s technically what PE AI products are supposed to do, studies using commercially available products and focusing on how AI complements radiologists (vs. comparing AI and rad accuracy) are still rare and worth a closer look.
The Diagnostic Study – A team from French telerad provider, IMADIS, analyzed AI and radiologist CTPA interpretations from patients with suspected PE (n = 1,202 patients), finding that:
- Aidoc PE achieved higher sensitivity (0.926 vs. 0.9 AUCs) and negative predictive value (0.986 vs. 0.981 AUCs)
- Radiologists achieved higher specificity (0.991 vs. 0.958 AUCs), positive predictive value (0.95 vs. 0.804 AUCs), and accuracy (0.977 vs. 0.953 AUCs)
- The AI tool flagged 219 suspicious PEs, with 176 true positives, including 19 cases that were missed by radiologists
- The radiologists detected 180 suspicious PEs, with 171 true positives, including 14 cases that were missed by AI
- Aidoc PE would have helped IMADIS catch 285 misdiagnosed PE cases in 2020 based on the above AI-only PE detection ratio (19 per 1,202 patients)
The Radiologist Survey – Nine months after IMADIS implemented Aidoc PE, a survey of its radiologists (n = 79) and a comparison versus its pre-implementation PE CTPAs revealed that:
- 72% of radiologists believed Aidoc PE improved their diagnostic confidence and comfort
- 52% of radiologists the said the AI solution didn’t impact their interpretation times
- 14% indicated that Aidoc PE reduced interpretation times
- 34% of radiologists believed the AI tool added time to their workflow
- The solution actually increased interpretation times by an average of 7.2% (+1:03 minutes)
The Takeaway
Now that we’re getting better at not obsessing over AI replacing humans, this is a solid example of how AI can complement radiologists by helping them catch more PE cases and make more confident diagnoses. Some radiologists might be concerned with false positives and added interpretation times, but the authors noted that AI’s PE detection advantages (and the risks of missed PEs) outweigh these potential tradeoffs.
|
|
|
Creating Your AI Platform Strategy
Adopting a platform strategy can simplify the deployment and management of imaging applications and AI algorithms, but there’s a lot to consider. In this eBook, Blackford Analysis and its clients detail how AI platforms can benefit clinical and IT teams, and share guidelines to consider when selecting a platform.
|
|
Nanox AI’s CPT III Codes
The American Medical Association recently added new CPT III codes for quantitative CT tissue characterization, paving the way for more health systems to adopt Nanox AI’s HealthCCSng CAC scoring population health solution.
|
|
- Clinicians of the Future: A survey of nearly 3,000 clinicians from Elsevier Health and Ipsos revealed that 70% of clinicians believe digital health technologies will enable a positive healthcare transformation and 59% believe that most clinical decisions will be based on AI-enabled tools. However, these technology shifts will face plenty of challenges, as 69% reported being overwhelmed with the current volume of data, 69% predict that digital health technologies will become a greater burden in the future, and 83% believe clinician training needs to be overhauled so they can keep up with technological advancements.
- HALO Dx & Precision Imaging’s JV: HALO Diagnostics launched a joint venture with Northeast Florida radiology practice Precision Imaging Centers, opening four locations that combine HALO Dx’s multi-modal diagnostic approach (imaging + genomics + biopsy + labs) with Precision Imaging’s local staff and relationships. This is one of the first major geographic expansions we’ve seen from HALO Dx since they emerged in 2019 after acquiring ten California imaging centers with a goal to become “one of the top radiogenomics networks in the United States.”
- Patient Safety Concerns: The Emergency Care Research Institute’s list of 2022’s top-10 patient safety concerns highlighted a number of major challenges that we’re also seeing within radiology, including staffing shortages (#1 concern), COVID’s effect on provider mental health (#2), racial biases in patient care (#3), and cognitive biases and diagnostic errors (#5). The report didn’t specifically mention radiology, but its recommendations focused on the entire healthcare organization, suggesting that radiology-focused initiatives or products that address these concerns might be more likely to receive organizational support.
- LiverMultiScan Coverage: Perspectum’s LiverMultiScan solution (quantifies liver tissue w/ mpMRI) hit another coverage milestone, after AIM Specialty Health deemed it “medically necessary for evaluating diffuse liver diseases,” paving the way for coverage by Anthem and BlueCross BlueShield. A number of Medicare Administrative Contractors also updated their schedules to cover LiverMultiScan, suggesting that its usage volumes are poised to increase significantly when its CPT codes go into effect in July (0648T and 0649T).
- The Value of Second Opinions: A new MD Anderson study highlighted the clinical value of performing second-opinion scans and interpretations when patients were originally scanned at outside institutions (that aren’t the #1 cancer centers in the US). Analysis of 915 outside abdominal CT and MRI studies revealed that 65% had worse image quality and 31% were inappropriate for oncologic management. Of 375 outside radiology reports, 34% (131) had interpretation discrepancies versus reports from MD Anderson subspecialists and 48% of those discrepancies (42 patients) led to treatment plan changes.
- WellSpan’s Aidoc Adoption: Pennsylvania’s WellSpan Health (200 locations, >1,600 physicians & APPs) announced its adoption of Aidoc’s imaging AI platform, starting with WellSpan York Hospital and expanding to other locations going forward. WellSpan will leverage a range of Aidoc abnormality detection and prioritization solutions (PE, ICH, c-spine fracture), aiming to help improve efficiencies, catch abnormalities that might have otherwise been missed, and expedite care coordination.
- Identifying False True Positives: The smart folks from CARPL.ai unveiled a new AI explainability validation technique that might help identify when AI algorithms make the right predictions for the wrong reasons. CARPL’s new Explainability Failure Ratio (EFR) process identifies when AI models localize part of an image that’s not within an explainability bounding box or near radiologist annotations, and then has a human expert determine if there is a logical reason for this failure. If there’s no logical reason, it’s likely that this is an Explainability Failure that should be flagged during the AI validation process.
- Edison Digital Health: GE Healthcare announced the upcoming launch of its Edison Digital Health Platform, which will enable healthcare systems to have a vendor-agnostic platform to integrate apps into clinical workflows. The platform will aggregate data from multiple sources (e.g. EMRs, labs, imaging, genomics) so that providers can easily access relevant patient information in a unified view, allowing them to break out different patient cohorts for analytics.
- LDCT + Incidental Programs: A new study in the Journal of Clinical Oncology highlighted the complementary role of LDCT lung cancer screening programs and lung nodule follow up programs. The prospective observational study monitored outcomes from concurrent LDCT and Lung Nodule programs (n = 5,659 & 15,461), finding comparable stage I/II and stage IV cancer detection rates (I/II 61% & 60%; IV 19% & 20%), while the LDCT program had higher three and five-year survival rates (3yr 80% vs. 64%; 5yr 76% vs. 60%). More notably, only 54% of the lung nodule program participants who were diagnosed with cancer would have qualified for LDCT screening based on current guidelines.
- Info Blocking Numbers: Nearly a year after the US info blocking rules went live (and radiology reports had to be made immediately available), 274 information blocking claims have been entered into the ONC’s online portal. Claims most often came from patients who were seeking access to their electronic health information (176 claims) and were either charged to see them or faced unnecessary delays. Although both of these are explicitly forbidden in the 21st Century Cures Act, surveys following the law’s implementation showed that there’s still widespread confusion on what information blocking actually entails.
- Annalise CXR Matches Rad CTs: A new study in The British Journal of Radiology found that Annalise.ai’s Annalise CXR deep learning algorithm identified abnormalities in chest X-rays comparably to radiologists interpreting chest CTs from the same patients (n = 1,404 trauma cases). The CXR AI tool performed better than radiologists’ CT interpretations in identifying pneumothorax (p = 0.007) and segmental collapse (p = 0.012), while radiologists performed better than AI for clavicle fracture (p = 0.002), humerus fracture (p < 0.0015), and scapula fracture (p = 0.014). Annalise CXR performed about the same as the radiologists in the identification of rib fractures and pneumomediastinum.
|
|
How to Evaluate Mammography Workflow
Women’s imaging has come a long way, but operational efficiency remains a challenge for many facilities. To help address this challenge, this Fujifilm post details the five questions women’s imaging facilities should ask when evaluating workflow management solutions.
|
|
- Join Microsoft and Nuance today (March 24th) at 2pm ET to learn how they are working together to develop Nuance’s next‑generation, AI‑powered diagnostic imaging platform.
- Riverain recently joined the exclusive group of AI vendors to receive Europe’s more-demanding Medical Device Regulations (MDR) certification, which requires healthcare AI products to attain higher risk classifications and provide far more validation evidence.
- Learn how new GE Healthcare MRI technologies are making PI-RADS v2.1 compliant prostate imaging at 1.5T “a game changer,” with improved image quality, shorter scan times, and a better overall experience for the patient.
- Hybrid care, digitization, and the move to the cloud are three of Intelerad’s Five Healthcare Trends to Watch detailed in this spot-on report.
- Have more echo studies than sonographers? See how Us2.ai was able to classify, segment, and annotate echocardiographic videos with similar accuracy as expert sonographers.
- Canon Medical is making its way through the US on its 2022 Mobile Tour, bringing its products and solutions directly to hospitals and providers in 50 US cities. Tune in to see when Canon is coming to you and watch highlights from its tour stops along the way.
|
|
|
|
|
