perm filename BIOMED[DOC,TOB]1 blob sn#157476 filedate 1975-05-01 generic text, type C, neo UTF8
COMMENT āŠ—   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002		Mendelssohn
C00006 ENDMK
CāŠ—;
	Mendelssohn
ADVANTAGES OF COMPUTER IMAGING
speed
expense
larger statistics
direct perception (ultraviolet)
repeatibility-fatigue, bias, ignorance
above human evaluation--male,female cells, urine cells example


most of the work has clustered into several categories, the classical
problems: other groups have a big lead here
I  CYTOLOGY cells
ASILOMAR MEETING - Leon Wheeless, Univ of Rochester  June 8-13
Automatic Cytology--Enginering Foundation
Volume from Asilomar 1 year ago
Journal of Cytochem, Vol 22, July 1974
1. chromosomes--payoff not in image processing
  a. metaphase studies--karyotyping--optical density, count them
    this is an intellectual exercise now
  b. banding--imaging techniques are starting to catch up with this
  c. flow techniques-break up cells-laminar flow
    examine fragments HERZENBERG(stanford)
   laser beam excites flourescent spectrum, precise measurements
    1000 objects/sec
    WILL LOOK FOR TWO CENTROMERES, find very low rates of aberrations
    Livermore is doing this work
2. other flow system applications
    Los Alamos, Livermore
  two stains-nucleus, cytoplasm, measure time of flight
    get relative dimensions
  scatter signal-texture, size, shape
    scattering vs angle in dozen channels
3. white blood cells/red
  several commercial systems-vidicon w/ moving stage
    Corning Glass, Geometric Data(Smith, Kline, French)
    they work rapidly at finding, primitively at characterizing
      where is nucleus, how heavy, color, etc
  flow systems (Technicon, Biophysics, Inc)
  scatter signal (Bloch Associates)


3. Pap smear, sputum, urine (exfoliated cells)  many groups, rich area
  flow systems--rare events
  scanning systems
  George Wied (Univ of Chicago)
    Peter Bartles, Univ of Arizona(image processing)




II TISSUES--NEXT LEVEL OF ORGANIZATION
1. Liver Sections (MIT) stopped
2. Bladder Tumor Sections (Prewitt, NIH)
scene analysis in sections is very difficult
  uncertain boundaries
  sections through complex 3d object




III MACROSCOPIC
fingerprints-law enforcement
face recognition-Bell Labs, picturephone
dermatological
radiographic
  UCDavis, Mayo Clinic
  evaluating lungs-miners' lungs for compensation
    USC project being broken up, people available
  shape of heart
  bone tumors--Univ of Missouri (Lodwick)

IV 3D RECONSTRUCTION
  central nervous system
  retina
  electron microscope
  radiographic images of machines
    Mike Eckstrom Livermore