acerjio

For MRI access to the raw data alone is probably not that helpful, with the exception of some of the neural network based upsampling approaches. The real power from working with k-space comes from developing custom pulse sequences.

The Cartesian (rectilinear) approaches to sampling k-space are generally very inefficient acquisition time-wise, but very tractable for a cheap computer to reconstruct via inverse FFTs. Non-Cartesian readouts (such as radial, spiral, etc) can radically reduce acquisition time (read: time for patients to lie in the scanner) but radically increase the computational complexity of the reconstruction. Estimates range from 1000x to 10000x times as complex if you use all the fancy stuff to maximize image quality and reduce acquisition time as much as possible.

There are numerous (solvable) technical challenges to implementing these strategies, but the biggest is financial. If one MRI scanner can do the work of 3 or 4 MRI scanners by moving patients through faster, that's a lot of lost revenue for manufacturers, so there's not much of an incentive to change the current paradigm.

A lot of the problem comes from the use of generative neural networks. If the prior is that the reconstructed images should "look" a certain way, then the algorithm will favor that. Some of our colleagues did early work with DL and got scared off of generative models due to finding issues with nonphysical results (read: broken layers of cortex in the brain, completely non-physical anatomy) that these models can generate from the undersampled raw data.

That said, there are other great ways to incorporate DL into MRI other than recon. I'm more interested in the use of DL for image segmentation, feature detection, potentially denoising, or other techniques on the image processing side. Those make a lot more sense as "top down" tasks that are well suited for neural networks.

This report highlights a big part of the problem in supplying physicians in the US. The issues with funding are captured well, as are the issues with the unmatched, that is medical school graduates who cannot secure a position and continue their training.

However, there are some important aspects of the modern residency system that are not captured well, that do contribute to the pipeline bottlenecks.

One major issue surrounds the UME - GME split, that is undergraduate medical education (that is medical students) vs. graduate medical education (that is residents). As a medical student now, one of the most striking comments is how our professors, especially those who are older, comment on how much they did during medical school. Right now, medical students learn under such close supervision most of the experience is shadowing-based, i.e. watching attending physicians, resident physicians, and midlevel providers. This means that on graduation as MDs, new interns have very limited experience in managing patients.

In turn, that makes intern year critical in training physicians. However, much of what is learned in intern year is often a retrenchment of what should have been covered in medical school by accreditation standards. The concerns about medical students graduating without a full set of the expected competencies have lead to the development of the Core Entrustable Professional Activities (EPAs), to create a framework by which students graduate with documented competency in each area.

Unfortunately, initial experiences with the EPAs have been mixed. This summer, an academic article reviewing schools experiences implementing the EPAs revealed that while most EPAs were being adopted well, some of the Core EPAs, specifically involving basic procedural competence (think drawing blood, starting IVs, basic bedside procedures), were considered more appropriate for GME [1], which adds to the GME curriculum in residency.

Given that the 4th year of medical school for most students consists of "elective time" and time off to interview for residency, shifting training from medical school to residency only adds to stress on the pipeline. For example, why should internal medicine residency be 3 years long? If graduates were documented as more competent on graduation, would a two year residency for primary care or prior to specialized subspecialty training be sufficient? In my experience working with residents at my medical school, 3rd year residents are in supervisory positions, dealing with less patient care than first and second year residents, the rest of their time seems to be consumed with applying for fellowship and studying for their Board certifications.

Of course, some of the dilution of undergraduate medical education comes as a result of defensive medicine, scope creep from midlevel providers, and excessive regulatory requirements. (Just look at all of the regulatory burden and turmoil around how attending physicians can attest medical student notes for CMS billing purposes. If anyone wants a migraine, I invite them to look into it.) Additionally, other causes of the dilution comes from the growth in US IMGs, mostly from the Caribbean medical schools. Those schools often pay to place students at US non-academic (often for-profit) hospitals that are less interested in the educational mission as they are in easy money from precepting students. In those cases, those students often have extremely circumscribed roles in the clinic, largely due to the institutional lack of emphasis and commitment to teaching [2].

Fixing the pipeline will require creating more residency positions, but it will also likely require changes all along the pipeline, such as making UME more meaningful and graduating MDs (or DOs) who are more ready for residency on Day 1, and then a serious look at residency length to ensure that GME experiences are meaningful and not simply "cheap labor [3]."

[1] https://journals.lww.com/academicmedicine/Fulltext/2021/0700...

[2] One of my preceptors who graduated from a Caribbean school flatly said that medical school was just for observation, and that she didn't actually do anything until residency. While she became an excellent physician, this attitude did not correspond to the educational expectations of our institution, nor do I think reflects the expectations of most of the US system.

[3] PGY (post graduate year) salaries are very, very low on an hourly basis. PGY-1 salaries are anywhere from $58k/year to $65k/year for FY22. However, as residents like to point out, when you calculate their salary across the hours they work, the rate is usually less than federal minimum wage.