Blood cell analysis is routinely used for aiding in diagnosis of disease. One widespread technique is that used by the Coulter Counter (Coulter Electronics), which measures the electrical properties of individual cells in a solution, for counting and sizing. Currently, the output used is the total blood cell count and the mean corpuscular volume. However, the output can readily be made available as a discrete histogram of 100 channels of relative sizes vs frequency; in other words, the relative cell volume distrrbution is given as a sequence of 100 numbers. In disease various changes can occur in this curve. For example, the relative proportion of certain blood cells may vary, or the size of the blood cells may change. The large majority of tests are normal, so that an automated technique for identifying potentially diseased patients would be useful. The method described establishes a master curve by adding, channel by channel, curves from normal blood donors. New, unclassified, histograms are then compared with the master curve. The x2 statistic is used to compare the unclassified histograms with the master curve. Confidence levels can be assigned to the classification, but only in relative terms. The x2 test treats all the channels in an identical fashion. By assigning weights to channels of greater importance, improvements in the classification result. Confidence levels can also be derived for the weighted x2 statistic and again used in relative terms.
The analysis of blood cell populations, both red and white, is routinely used as an aid in the diagnosis of diseases. There are basically two methods, visual inspection and sizing of cells in a liquid suspension. In disease the relative proportion of different cells, their size, and their shape can change. A brief review of current methods for obtaining diagnostically useful indicators of disease from blood cell inspection is given in the following section. In this paper, the relative volume (size) distributions (histograms) of blood cells is measured by the Coulter Counter (4), which is also described in the following section.
Typical histograms of red and white blood cell size are shown in Section 3. A wide variation in these histograms is noted. In practice, a large majority of blood cell tests are normal. What is required, therefore, is an automated means for screening out only potentially diseased patients. A microcomputer interfaced to the Coulter Counter could perform this function. 199