Basically Bone -- A Keyword Searchable Image Database

• Thus, a radiologist presented with an "unknown" x-ray, compares its appearance with similar ones he/she has seen in the past.

• This comparison gives rise to a list of possible diagnoses, which can be narrowed by knowing the patient's clinical history, presenting complaint, the result of various laboratory tests, etc.

If we were able to develop such databases on the Amiga computer, we would have a "peripheral brain" we could consult anytime we had a challenging x-ray -- and they ALL are challenging! We could use such a database in several ways:

1. If we had an x-ray of the spine which showed a white, or "sclerotic," vertebra, we much search under "adult," "spine," and "sclerosis" to look for a similar image.

2. If we thought cancer were involved, we could look at images corresponding to "adult," "spine," and "cancer" to find a matching pattern.

3. We could answer questions from our medical colleagues as to what a specific disease in the spine might look like.

4. We could teach and quiz our Residents (physicians-in- training) regarding the radiologic appearance of disease in the body.

There are many picture databases available for the Amiga that will make and display small "thumbnail" pictures for easy retrieval of specific images from a disk filled with images. However, none, with the exception of GraphicRecall (Focus GbR), allows both sophisticated, multiple keyword search capabilities as well as the ability to annotate each image in detail.

• Thus, an x-ray demonstrating tuberculous of the spine can include an extensive description of this disease.

• The visibility of this information is under user control There are several advantages:
  • Text need not be a distraction while viewing images.

  • Images can be used as a "test question" with the write-up serving as "answers."

• GraphicRecall allows the write-up to be printed out, thus increasing its utility still further!

So, let's see how to go about creating a keyword searchable image database.

1. The first step is to assemble a series digital radiographs.
   • Digital images can be obtained in many ways. Some of these are discussed in the Photo Nukes section.

   • It's best to put each type of study on its own disk -- Chest studies on one, Bone on another, Gastro-Intestinal on a third, and so forth.
     • Removable disks. e.g. Bernoulli, Syquest, are appropriate here.

2. The next step is to ask GraphicRecall to make thumbnails of each image.
   • You first must specify the size and color depth of the icons you wish GraphicRecall to create.

   • You must then tell GraphicRecall which images you want thumbnails for.

   • GraphicRecall will then launch Art Department Professional (ADPro, Elastic Reality) which will batch-processes the list.

3. Now comes the hard part. You must describe, in detail, the disease that corresponds to each image including --
   • Which ages are most commonly affected.

   • How does the patient appear clinically.

   • What are the common x-ray findings.

   • What other diseases might have similar x-rays.

   • How is the disease treated.

   • What is the prognosis.

4. You then need to add keywords to each image so that the relevant x-ray can be found during a search.
   • For example, searching under "popcorn calcification" leads to x-rays of "enchondroma" and "chondrosarcoma."

   • The user may then display these images to see which one most resembles his/her "unknown" x-ray.
     • An enchondroma might (with a fracture) look like:

     • A chondrosarcoma might look like:

   • If a keyword search provides too many possibilities, Graphic RECALL allows the user to do a keyword search of the diseases in the list. What a powerful feature!

The "Basically Bone" database can be used in other ways.

• Suppose you wanted to know about a disease called "acro-osteolysis."
  • By using "acro-osteolysis" as a keyword, you can locate a radiograph of this disease and read something about it as well.

    

  • As another example, suppose you wanted to quiz a student about a certain bone disease.
    • You could display a radiograph of the entity --

      

    • Ask the student to discuss the case --
    • Allow the student to see the case write-up.

    • Print out the write-up so that the student can study the disease at leisure.

  • Finally, suppose a student wanted to review his/her knowledge of Bone radiology.
    • He/she could "work through" the set of images, first viewing them and then reading the write-ups.

    • Alternatively, two students could quiz each other.

So, what is so great about "Basically Bone?"

• It can be updated continually.
  • Images can always be added.

  • Text can always be revised.

  • There are no size limitations.

• Search facilities are excellent and fast.

• "Basically Bone" is fast -- information is literally at your fingertips.

• "Basically Bone" is practical -- a real adjunct in training and diagnosis.

• Put everything on a CD ROM and it becomes readily distributable.

• Best of all, by writing up each case, one actually reviews -- or learns! -- a lot of bone radiology!

There are certain advanced features of Graphic Recall utilized by "Basically Bone." Unlike many databases,

• One can simultaneouslydisplay several images corresponding to the same search word, e.g., osteosarcoma --

  

• OR, display different images corresponding to different search words, e.g. to see the difference between enchondroma and chondrosarcoma --

  

So, what is the next step?

• It would be possible to set up key word searchable databases for Chest, Gastro-Intestinal, etc. images similar to "Basically Bone." These could include CT, MR, pathology, etc.

• A CD-ROM is always an option.

• Porting the functions of GraphicRecall to the World Wide Web to create an Internet searchable database is also enticing.

Realistically, I am only a single individual working on these rather complex projects. At best I can only show what can be done and must leave it to large, well-funded groups to carry out the major effort of implementation.

revised -- January, 2000
revised -- December, 2002

List of Projects -->