Megapixel myth

| March 3, 2008

A site created by a photographer, Ken Rockwell at www.kenrockwell.com addresses a variety of issues dealing with digital and film photography, image processing and helpful tips.  Among them are buying guides, technical support and well done photo demonstrations of white balance, sharpening and instrument recommendations. 

Included is a page dealing with the "megapixel myth" which comes up for static and telepathology pathology images when trying to answer the questions "How much resolution is enough?"  "Are the images big enough?"  "What size camera should I buy?" And so forth. 

One issue I come back to frequently is that resolution and little to do with image quality.  Other factors including subject matter, field selection, stain quality, how the image will be viewed, what is/are the questions being asked or what illustrations are provided for are much more important. 

Realize that video telepathology may be sufficient for most cases with a resolution of less than 640×480.  Static telepathology is accomplished with images of that size without compromising the diagnostic features or the quality of the image if the appropriate fields and magnifications are selected.  Robotic systems allow for the same or slightly more with minimal gain in my experience. 

The monitor resolution or DPI obviously is critical.  With that said, I have seen little lost other than desktop space with bulky CRT monitors without flat panel higher resolution screens for diagnostic work. 

Breastca1This breast cancer image was captured with a 3 MP camera (640×480), originally saved as a TIFF, compressed to a JPEG and further compressed for this posting at a size of less than 300K.  It represents a diagnostic field.  I have used "better" cameras with inferior image quality. 

Mal7This time stamped photo of malaria taken from a thin blood film using an oil immersion lens was taken using a 3 MP camera through the ocular of a BH-2 microscope.  You can make out the organisms and speciate accordingly.  Not bad considering the "technology".

Some publishers still insist upon "high resolution" iamges, saved as TIFF for images accompanying manuscripts.  One I recently submitted a paper to required RBG and CMYK formats before deciding if and what would be published in the mode of their choice.  Absurd given the image size, paper and print quality.  It speaks to resolution (linear resolution) in terms of image resolution, print resolution and the issue of screen resolution I touched on above.  There is little difference between 3 and 6 MP.  The difference is nil between 6 and 8 MP.  One needs about a doubling of resolution or film size to make an obvious improvement.  This is the same as quadrupling of megapixels. 

For a print in a journal at column width printed at 300 DPI, the degree of resolution is not as significant as focus, tone, white balance, contrast, sharpness, etc…

Dsc_2989Dsc_2990The image on the far left was taken with a 6 MP camera, the one on the near left taken minutes later with an 8 MP camera.  Can you tell the difference (other than slight differences during sunrise)? 

20x1adobe

20x1adobe2

20x1adobe4The same is true of histology images.  Each of these was taken at 20x with a 2MP camera and compressed so that each one is 1/2 the size of the other from left to right in this pane. 

All are from the same field of a Warthins tumor.  Not only can you make the diagnosis in the most compressed image, there is no difference between the 3 that would or should dissuade one from making the same diagnosis.

I fully realize there are issues with the blog software but the point is the difference is nil regardless.

With that said, I am looking at personal cameras beyond 8 MP now and into double digits (a complete overkill for 5×7 prints, even 8×10 or larger posters).  Part of it is to try to get closer to "film" quality which requires resolution beyond most commercially available cameras. 

For more on this topic, check out Ken Rockwell’s discussion on the topic


Category: Digital Pathology News, Histology, Microscopy

Comments are closed.