Week 5

It’s been a productive week from my project standpoint – chugging away at 700 sets of patient CTs (if I ever see a CT after this term is over, it had better be my own)!  Data collection is cumbersome, but I’m hoping by Monday it’s all out of the way to start with the fun part – analysis.

To elaborate a bit more on my project, we’re looking to see if there are certain physical parameters associated with how the lung biopsy is performed that affect the rate of pneumothorax. Why is that important? Let’s talk through an example together.

Pictured below is the CT-image right at the point of biopsy of the legion (circled in blue below, needle shows up as white on contrast). While there are general rules and trains of thought with regards to how the biopsy should be performed based on past data (see the referenced review [1]), understanding as to the influence of skin-pleural tissue distance, pleural tissue traversed, density of the pleural tissue, size of the lesion, and angle of incidence affect post-operative outcome is not well understood. The clinician selects a specific path and its associated parameters, but an infinitely dense n-dimensional parameter space exists (illustrated poorly by the triangle that I’ve added below).

As I’ve mentioned previously, I’ve be fitting these parameters to a logistic regression model. That means someone has to go extract those 5 previously mentioned parameters (as well as some other information as to how the surgery is performed). I can do around 20 of these an hour, so it’s a long process, but this type of data is worth it.

One other small note about the parameters is the density, measured in Houndsfield units (HU). The software that I have access to automatically calculates it for me, but that doesn’t mean I shouldn’t understand how it’s calculated. Each pixel has an associated value, as a function of contrast (scaled in these cases against air, at -1000). The Houndsfield unit simple transforms the scaling to be against water, such that each unit increase is 0.1% against the value of water. Lung tissue is closer ~-600 range, given that a large majority of the lung is air!

Finally, being healthier than I was last week, I was finally able to attend rounds in the mornings like I had been planning. Truth be told, it was another moment of being in total awe at the body of knowledge that these clinicians have. They were able to go through 10 cases in approximately 30 minutes and discuss any potential modifications they would have made to the past treatment as well as potential treatment options moving forward. My experiences in surgery for about two weeks definitely helped with the vocabulary and making sure I wasn’t lost, as I had actually seen about half of the cases they were discussing!

All in all, as this experience starts to wrap up, I’m more and more excited about the type of data we’re getting. Hopefully I’ll be able to update on some research success next week!

1 - Winokur RS, Pua BB, Sullivan BW, Madoff DC. Percutaneous lung biopsy: technique, efficacy, and complications. Semin Intervent Radiol. 2013;30(2):121–127.