Apparatus for identification and integration of multiple cell patterns

Claim 3

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the single cell classifier identifies and classifies single cells using a single cell classification process.

Claim 4

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the single cell classifier output includes a number of alarms and a percent of negative objects among all detected objects in a field of view.

Claim 5

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the group classifier detects and classifies groups of cells formed in sheet or syncytium arrangements using a group classification process.

Claim 6

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the group classifier output includes a number of detected groups, a percent of abnormal groups among all detected groups in a field of view and a percent of normal groups among all detected groups in the field of view.

Claim 7

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the thick group classifier detects and classifies thick groups of cells formed in three dimensional clusters of cells using a thick group classification process.

Claim 8

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the thick group classifier output includes a number of detected thick groups and a percent of abnormal thick groups among all detected thick groups in a field of view.

Claim 9

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the integration means further provides single cell results, group results, thick group results, field of view integration results, and contextual information components.

Claim 10

The apparatus for identification and integration of multiple cell patterns of claim 1 further including a means for detecting when the plurality of classifier outputs are inconsistent, connected to the plurality of classified outputs having an inconsistency output.

Claim 11

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the single cell classifier output indicates that a number of single cell alarms is greater than a first amount and the group classifier output indicates that a percent of groups classified as abnormal is less than or equal to a second amount and a percent of groups classified as normal is greater than a third amount.

Claim 12

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the single cell classifier output indicates that a number of single cell alarms is less than or equal to a first amount and the number of single cell negatives is greater than a second amount and the group classifier output indicates that a percent of groups classified as abnormal is greater than a third amount.

Claim 13

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the single cell classifier output indicates that a number of single cell alarms is greater than a first predetermined amount and a thick group classifier output indicates that a percent of thick groups classified as abnormal is less than or equal to a second amount and a number of thick groups is greater than a third amount.

Claim 14

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the single cell classifier output indicates that a number of single cell alarms is less than or equal to a first amount and a number of single cell negatives is greater than a second amount and the thick group classifier output indicates a percent of thick groups classified as abnormal is greater than a third amount.

Claim 15

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the group classifier output indicates a percent of groups classified as abnormal is greater than a first amount and the single cell classifier output indicates a number of single cell alarms is less than or equal to a second amount and a number of single cell negatives is greater than a third amount.

Claim 16

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the group classifier output indicates a percent of groups classified as abnormal is less than or equal to a first amount and a percent of groups classified as normal is greater than a second amount and the single cell classifier output indicates a number of single cell alarms is greater than a third amount.

Claim 17

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the group classifier output indicates a percent of groups classified as normal is greater than a first amount and the thick group classifier output indicates a percent of thick groups classified as abnormal is less than or equal to a second amount and a number of thick groups is greater than a third amount.

Claim 18

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the group classifier output indicates a percent of groups classified as abnormal is less than or equal to a first amount and a percent of groups classified as normal is greater than a second amount and the thick group classifier output indicates a percent of thick groups classified as abnormal is greater than a third amount.

Claim 19

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the thick group classifier output indicates a percent of thick groups classified as abnormal is greater than a first amount and the single cell classifier output indicates a number of single cell alarms is less than or equal to a second amount and a number of single cell negatives is greater than a third amount.

Claim 20

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the thick group classifier output indicates a percent of thick groups classified as abnormal is less than or equal to a first amount and a number of thick groups is greater than a second amount and the single cell classifier output indicates a number of single cell alarms is greater than a third amount.

Claim 21

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the thick group classifier output indicates a percent of thick groups classified as abnormal is greater than a first amount and a percent of groups classified as abnormal is less than or equal to a second amount and the group classifier output indicates a percent of groups classified as normal is greater than a third amount.

Claim 22

The apparatus for identification and integration of multiple cell patterns of claim 9 wherein the means for detecting when the plurality of classifier outputs are inconsistent further comprises a means for detecting if the thick group classifier output indicates a percent of thick groups classified as abnormal is less than or equal to a first amount and a number of thick groups is greater than a second amount and the group classifier output indicates a percent of groups classified as abnormal is greater than a third amount.

Claim 23

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the apparatus further comprises a means for reclassifying the multiple cell patterns having an input connected to the single cell classifier output, the group classifier output, and the thick group classifier output.

Claim 24

The apparatus for identification and integration of multiple cell patterns of claim 22 wherein the means for reclassification further comprises a means for reapplying the plurality of classifiers in an order based upon a predetermined goal to provide a final outcome.

Claim 25

The apparatus for identification and integration of multiple cell patterns of claim 23 wherein a difference between the at least one output and the final outcome for each field of view is accumulated for the biological specimen and used for slide classification.

Claim 26

The apparatus for identification and integration of multiple cell patterns of claim 23 wherein the predetermined goal is classification of abnormal vs. normal conditions and an inconsistent condition resolution sequence comprises: a) adjusting the thick group classifier conditioned on the single cell classifier output to provide an adjusted thick group classifier output; b) adjusting the group classifier conditioned on the adjusted thick group classifier output to provide an adjusted group classifier output; and c) adjusting the single cell classifier conditioned on the adjusted group classifier output.

Claim 27

The apparatus for identification and integration of multiple cell patterns of claim 23 wherein the predetermined goal is classification of endocervical cell conditions and a inconsistent condition resolution sequence comprises: a) adjusting the thick group classifier conditioned on group classifier output to provide an adjusted thick group classifier output; b) adjusting the single cell classifier conditioned on the adjusted thick group classifier output to provide an adjusted single cell classifier output; and c) adjusting the group classifier conditioned on the adjusted single cell classifier output.

Claim 28

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the integration means provides a feedback signal to the plurality of classifiers.

Claim 29

The apparatus for identification and integration of multiple cell patterns of claim 1 further including a means for determining whether the plurality of classifiers may be reapplied to the image output having an input connected to the means for integration.

Claim 30

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the plurality of integrated outputs indicate whether the biological specimen is normal or requires review.

Claim 31

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the plurality of classifiers provide at least one classified output indicating likelihood of abnormality.

Claim 32

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the biological specimen is prepared by the Papanicolaou method.

Claim 33

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein the biological specimen is a gynecological specimen.

Claim 34

The apparatus for identification and integration of multiple cell patterns of claim 1 wherein classified outputs for the plurality of classifiers are correlated and accumulated over a biological specimen slide.

Claim 35

The apparatus for identification and integration of multiple cell patterns of claim 33 wherein correlation of field of view results for the plurality of classifiers includes a check selected from the group consisting of a first co-occurrence frequency including a single cell alarm having at least a group alarm in the same field of view, a second co-occurrence frequency including a single cell alarm having at least a thick group alarm in the same field of view, a third co-occurrence frequency including a single cell alarm having at least a group alarm and at least a thick group alarm in the same field of view, a fourth co-occurrence frequency including a single cell alarm having no thick group alarm yet having at least one group normal object in the same field of view, a fourth co-occurrence frequency including a single cell alarm having thick no group alarm yet having at least one thick group normal object in the same field of view, a sixth co-occurrence frequency of a single cell alarm having no group or thick group alarms and yet having at least a group and a thick group normal object in the same field of view, a seventh co-occurrence frequency including a group alarm having at least a single cell alarm in the same field of view, an eighth co-occurrence frequency including a group alarm having at least a thick group alarm in the same field of view, a ninth co-occurrence frequency including a group alarm having at least a single cell alarm and at least a thick group alarm in the same field of view, a tenth co-occurrence frequency including a group alarm having no single cell alarm yet having at least one single cell normal object in the same field of view, an eleventh co-occurrence frequency including a group alarm having no thick group alarm yet having at least one thick group normal object in the same field of view, a twelfth co-occurrence frequency including a group alarm having no single cell or thick group alarms and yet having at least a single cell and a thick group normal object in the same field of view, a thirteenth co-occurrence frequency including a thick group alarm having at least a single cell alarm in the same field of view, a fourteenth co-occurrence frequency including a thick group alarm having at least a group alarm in the same field of view, a fifteenth co-occurrence frequency including a thick group alarm having at least a single cell alarm and at least a group alarm in the same field of view, a sixteenth co-occurrence frequency including a thick group alarm having no single cell alarm yet having at least one single cell normal object in the same field of view, a seventeenth co-occurrence frequency including a thick group alarm having no group alarm yet having at least one group normal object in the same field of view, and an eighteenth co-occurrence frequency including a thick group alarm having no single cell or group alarms and yet having at least a single cell and a group normal object in the same field of view.

Claim 38

The method of claim 35 wherein the single cell classifier detects and classifies free-lying cells and non-nuclear overlapping cells using a single cell classification process.

Claim 39

The method of claim 35 wherein the group classifier detects and classifies groups of cells formed in sheet or syncytium arrangements using a group classification process.

Claim 40

The method of claim 35 wherein the thick group classifier detects and classifies thick groups of cells formed in three dimensional clusters of cells using a thick group classification process.

Claim 41

The method of claim 35 further including the step of providing single cell results, group results, thick group results, field of view integration results, and contextual information components.

Claim 42

The method of claim 35 further including the step of indicating whether the biological specimen is normal or requires review.

Claim 43

The method of claim 35 further including the step of generating a likelihood of abnormality.

Claim 44

The method of claim 35 wherein the biological specimen is prepared by the Papanicolaou method.

Claim 45

The method of claim 35 wherein the biological specimen is a gynecological specimen.