Personal Notes and Observations from RSNA 97

This is the first RSNA that I've been able to go to. So, as a newcomer to the scene, I thought I might share some of my experiences at the biggest radiology related conference in the world.
Sat Nov 29 Sun Nov 30 Mon Dec 1 Tues Dec 2 Wed Dec 3 Thu Dec 4 Fri Dec 5

November 29, 1997 - Day 1

This year, it was in Chicago, IL (as usual) from Nov 29 - Dec 5, 1997. After registering and going through the program guide picking out all the technical and scientific exhibits, scientific sessions, and courses I wanted to go to, I quickly discovered that I would need 10 other copies of me just to see even half of what was here. Fortunately, I was warned about this, but even still, I was surprised at how many interesting things there were that I wanted to see.

I arrived in Chicago on the 29th, which seems to be kind of a pre-game warmup day. The exhibit halls aren't scheduled to open until the 30th, so most of the booths were still in the middle of being constructed. From what I got to see as I wandered around a bit, one could spend an entire week just browsing among the booths. No big crowds just yet.

I'm quite impressed by downtown Chicago. Especially Michigan Ave, also known as the Magnificent Mile. Quite the shopping mecca. And all sorts of interesting looking restaurants to check out. One in particular, the Rainforest Cafe looks like a lot of fun. The outside is decorated up to look like a rainforest (surprise surprise), and I'm told the interior is similarly decorated. I think I'll have to check it out sometime this week.

Sunday, November 30, 1997

This was the first real day of the RSNA. First stop was the Scientific Exhibit/InfoRad Hall.

The first exhibits I encountered was 0392PH and 0394PH. Fascinating hologram-like 3-D images using an array of cylindrical lenses. Just like those magic images that you tilt back and forth to make the picture change. A little more sophisticated. According to the authors, up to 61 images can be recorded on a single film. Move side to side and watch a cerebral CT angiogram rotate side to side. In another image, move up and down and watch a cine loop of a cardiac MUGA scan.

At the InfoRad theatre, I caught a presentation titled Measuring and Maintaining Image Quality on Medical Image Displays. This was an interesting presentation on simple test objects that can be used to quickly test the image quality of display workstations on a daily basis. The bottom line was that such tests should be done daily, using a simple quick test that the radiologist can perform before starting into the cases of the day. Results should be automatically logged and tracked.

Wandering aroud the technical exhibits can be somewhat overwhelming. So many vendors exhibiting all sorts of products. At the Nuclear Associates booth, the most interesting item was the new CR chest phantom developed by people at Duke University. It's not quite a shipping product just yet, but a mock up and radiograph of the phantom were available to look at.

Over at Agfa, their new CR equipment looks quite impressive. The CR unit itself is compact and autofeeding. The cassettes are stacked on edge on one side of the CR unit. The reader pulls the cassette in, reads it and spits it out on the other side of the reader. And it's surprisingly small, just over 1 m tall. The QC workstation is likewise impressive. Pentium based, Agfa has an impressive collimation detection routine. Both Fuji and Kodak CR systems seem to have occasional difficulties with recognizing the collimator edges. The Agfa system seems to much better in this respect, even allowing 2 radiographs on one cassette, something you can't do with the Fuji or Kodak systems. Even more impressive is Agfa's MUSICA image processing software. MUSICA applies unsharp masking to obtain unsharp mask images at several frequencies. Applying different weights to the unsharp masks and applying them to the original image produces impressive results. Definitely worth a closer look.

One other thing I learned was to leave before 4:30. That's when the long lines for the shuttle buses start building up. At the South building that is anyway. I've heard that the lines are much shorter over at the North building.

Monday, December 1, 1997

Today was a busy day. Two categorical course sessions, scientific sessions, vendor meetings and dinners. Somewhere in between there, I managed to squeeze in some more time at the scientific exhibits and InfoRad. And the shuttle bus lines are definitely much shorter over at the North building.

At the Sterling booth, their contribution to the DR (direct radiography) market was shown off. An impressive product. They've combined the X-ray generator console with the QC workstation. Fischer x-ray generators and equipment with a Sterling name slapped on it rounds out their x-ray room. Very promising looking. It seems to me that their image processing needs a little more development though.

In the afternoon, there was the Physics Scientific Session on Nuclear Medicine.

Paper 510, Comparison of Coincidence Imaging Parameters of a Dual-head Scintillation Camera with 3/8" and 5/8" Thick Crystals was an interesting one. The authors reported that the FWHM for high resolution collimators was 4.2 mm at 3/8" and 4.5 mm at 5/8". Slit collimators (like x-ray antiscatter grids) were used with a graded absorber made of lead, copper and tin to reduce scatter. I don't recall if there were spatial resolution results given for 140 keV photons.

Paper 512, Development of a Quantitative Imaging and MIRD Dosimetry Software Application for Radionuclide Therapy Treatment Planning Using a Fourth Generation Language (whew) uses IDL for determining the radiation dose to various organs based on single or opposing views. Interesting looking stuff.

In Paper 513, Combining X-ray and Radiotracer Mammography Images, a stereotactic breast biopsy and small PET imager are combined to improve localization of non-palpable lesions. Interesting concept using image segmentation to isolate lesions from the PET scan, and registering it to the x-ray image.

Tuesday, December 2, 1997

Today was slightly more relaxed. More PACS at GE's booth, lunch with Medasys, PACS stuff again with Imation and more InfoRad. Everybody's got PACS. Just about everywhere you went, someone was displaying workstations with monitors all over the place. Many booths were showing off big optical and CD jukebox archives too, with capacities ranging from a few hundred GB to over 3 TB (terabytes).

Next door to Medasys was a company called Digirad. There offering was a very interesting solid state camera with an 8"x8" detector size composed of 3 mm x 3 mm CZT (cadmium-zinc-telluride) detectors. Very impressive energy resolution and pretty decent sensitivity. It's a mobile unit, very compact and lightweight, so it will go anywhere. They also offer a SPECT version in the form of a rotating chair, with the idea being that you can quickly evaluate someone coming into an ER complaining of chest pains. I'm very intrigued.

The morning's Physics scientific session was on PACS and Workstations. All very interesting presentations.

Paper 701, Reliability of a Hospital Wide Picture Archival and Communication System (PACS) Using a Central Arhitecture: Analysis of the Four Year Experience at the Baltimore VA Medical Center was an interesting one. An examination of the uptime of the PACS system. A surprising finding was that the main source of system downtime was the CR reader. Even a 98% uptime was found to be unacceptable for clinical use. Scheduled downtime should be arranged to occur in the background whenever possible, so that the system does not need to be taken down at all.

Paper 705, Accuracy of Radiologists Using a Computer Workstation in Comparison to Film in the Interpretation of CT Examinations was quite good too. Even though radiologists were found to be 15% faster when reading softcopy CT, there were no statistically significant differences in accuracy.

Paper 707, Interpretation of Chest Radiographs on High-resolution PACS Monitor: Optimal Ambient Illumination in Reading Room studied the effect of ambient lighting on the accuracy of softcopy and lightbox reading of chest radiographs. The authors reported that reading time increased as the ambient lighting level increased. Eye fatigue became a problem when reading softcopy as lighting levels decreased. I think this is something everyone's experienced when working in front of a computer monitor for extended periods.

At the InfoRad exhibit, exhibit 9311NT: 100% Pure Java DICOM Paradigm showed off a very sophisticated looking DICOM image viewer, all implemented in Java. The authors plan on making their Java class library available for download sometime after the RSNA in the hopes of stimulating more related work.

Next door at exhibit 9312NT: Wavelet Compression Evaluation Tool for DICOM Images demonstrated an interesting tool for playing with wavelet compression and observing how it affects the compressed image. An interesting looking tool for demonstration and teaching purposes.

Wednesday, December 3, 1997

Midweek of RSNA. Many of those who have been here since Saturday or Sunday have left or will be leaving today, and a new batch of people are arriving.

The Categorical Course on the basic physics of MRI is quite good. I'm learning quite a bit from the lectures, and the speakers are very good too. The course syllabus ought to be a fairly decent read too (if you're into that sort of thing). At $45, it can be a bit of a bite into your budget though. I suspect it will be a good addition to my library though.

After today's installment of the MRI categorical course, it was off to another Scientific Session on Computed Radiography. More excellent presentations. One interesting one was Paper 1061, Contrast, Noise, and Patient Dose Compromises in Digital Chest Radiography. The authors examined the behaviour of digital chest radiographs to added filtration. They reported increased SNR, but slightly decreased contrast when a small amount of Mo filtration was added to a 110-120 kVp x-ray beam.

Paper 1065, Impact of Matrix Size and Processing in Storage Phosphor Chest Radiography: Results of an Anthropomorphic Phantom Study looked at the difference in image quality between 2k and 4k digital chest radiographs. They found no significant difference in lesion detectability with the 4k radiographs. The authors also looked at using a larger kernel size for the unsharp masking filter and found that a larger kernel size with slightly less edge enhancement improved lesion detection.

In Paper 1067, Image Quality and Dose in Digital Chest Radiography: Comparison of Standard and 150 kVp Techniques with and without an Antiscatter Grid, the authors reported that performing chest radiographs at 150 kVp resulted in a decrease in patient dose while producing a slightly better image and little difference in image noise. Additionally, images obtained at 1/4 the exposure used for film/screen were found to be acceptable for diagnostic purposes.

In the afternoon scientific session in CR, several presentations were repeats of the morning session. Paper 1253, Automatic Correction of Orientation of CR Chest PA Images presented an interesting way of automatically reorienting PA chest images. By locating the centroid of the region containing pixel values in the top 15% of the image, the chest image was rotated so that the location of the centroid would always be in the lower half of the image.

Today was a short day, so I took the opportunity to get off my feet for a while. After all the walking around I've been doing, it was a welcome break.

Thursday, December 4, 1997

Last day of the technical exhibits, so I spent a good portion of it wandering around the various booths. Primarily, I was looking at the nuclear offerings. GE has a nice compact little dual head camera. Much more compact that most of the other systems out there, without sacrificing much in the way of detector size. The gantry consists of a ring, much like the old GE Starcam single head systems. If they put it on wheels, they'd have a mobile SPECT system. I wonder if there would be much demand for that. I've also heard that their software is written in Visual Basic too.

Picker's AXIS and IRIS systems I thought were the most interesting developments. The detectors of the AXIS (triple head) and IRIS (dual head) not only rotate around the COR, but pivot and move tangentially to the orbit. Very cool. With their large detector areas, the cameras are sure to be extremely versatile.

Just about every vendor is offering coincidence imaging, with either 9.5 mm (3/8") or 15.9 mm (5/8") thickness crystals. SMV was also offering it with a 12.7 mm (1/2") thick crystal. Most of the images looked quite good. I think it's an area I'll have to start getting in to. An interesting note about SMV is that their workstation is running OS/2. This may be an issue for those wanting to keep the number of software platforms that need to be supported to a minimum. I was kind of surprised to discover how many vendors still have non-uniform attenuation correction as works-in-progress.

At the Inforad exhibits, I spent some time at exhibit 9301NT, Exploration of the Effects of Acquisition Methodology on Image Quality for a High Performance Diagnostic Active-matrix Flat-panel Imager (AMFPI). Very cool stuff. Good radiographic resolution. The pixels are on the order of 0.1 mm in size. With a fast readout, they've done fluoro too at 2 frames/sec. I wonder if anybody's tried these things for nuclear imaging.

In the scientific exhibits, I found several interesting poster displays on chest imaging. Over at exhibit 0829CH, CR chest x-rays at 150 kVp with heavy filtration (3.5 mm Al + 1.5 mm Cu) demonstrated improved lesion detection.

Publisher's Row is a fun place to browse around. Very busy too. Although most of the books were medical in nature, the MTMI booth had a fairly good selection of physics oriented books. I've been looking around for a good textbook on Nuclear Medicine physics. I found one book that covered the basics, but didn't go into as much detail as I was looking for. Sorensen and Phelps is still the standard text, but it's getting a bit dated. Maybe I should write my own...

Friday, December 5, 1997

Last day, finally. Whew, what a week. RSNA is quite the event, and if you're there the whole week, it requires a bit of pacing. I was starting to burn out from information overload by Wednesday. After the last session of the MRI categorical course, it was onto the shuttle bus and to the airport for the trip back.
Table of Contents
About me My Geek Code Research interests
My coin collection Neat Stuff Places I surf My PGP keys Contacting me
Eugene Mah - eugenem@ix.netcom.com
Last modified: Tues Oct 7 18:08:15 Eastern Daylight Time 2008