The Application Of Computer Aided Detection In Computerized Tomography Of Detection For Colon Diseases

Objective:Colon diseases are of the diseases that threat human beings'health and lives. So, it is of great importance to diagnose theses diseases early for the treatment and prognosis. The common methods we use now in clinical are occult blood test, barium enema examination, colonoscope examination as well as Computerized Tomography Colonography (CTC) which has developed in recent years. Filling the colon with low-density gas, the radiologists scan the subjects with spiral computerized tomography. The images are to be post-processed by computers CTC imaging post-processing includes Raysum images, Computerized Tomography Virtual Endoscopy(CTVE),Virtual Dissection (VD) for colon, Shaded Surface Display(SSD), Multiplanar Reformation (MPR). Computer Aided Detection(CAD) is auto-detect region of interesting by computer based on the density, volume and surface of the tissues so as to remind the doctors to pay attention to the change of these regions. What's more, CAD is capable to auto-detect, mark and show the lesions combining with CTVE. The research diagnoses colon diseases with CAD combining with CTC, and discuss the implication value.Methods: The research used 16 spiral Computerized Tomography or 16 spiral Computerized Tomography (Lightspeed, GE).The research examined 97 subjects that are supposed to have colon diseases by CTC from May 2006 to February 2009. There are 45 male subjects and 42 female subjects, aging from 17 to 77, and the average age is 46 years old. All the subjects are supposed to have intestinal tract cleaning, and intramuscular injection anisodamine 10-20 mg to the subjects with no contraindication(glaucoma, hypertension and prostatic hypertrophy use with caution) to decrease intestine tension and peristalsis. All the subjects are supposed to have left side-lying position and injected air 900-1000 ml by anus at first, and then have right side-lying position and injected air 800-1100 ml. The subjects are supposed to have the sense of expanding-gas, be judged by CTC and replenished air according to the situation. The colon should be expand with air and haustra coli showed clearly. CT scanned from diaphragm to low side of pelvic where all the colon should be air expanded. Parts of the subjects had enhanced CT dual-phase scanning, using 64 spiral computerized tomography (LightSpeed 64, GE). The enhanced drug iohexol(300mg I/ml, 90-100 ml) was injected through elbow vein with the speed of 3.5 ml/s.All the cases were diagnosed by four radiologists, two of which were expert readers and two were non-expert readers, thus they were divided into two groups. The non-expert group had the experience of cooperate with expert readers to diagnose colon diseases. The expert group had sufficient experience of diagnosing colon diseases and operating CAD software. The fore doctors are supposed to diagnose with CTC with/without CAD blinded and respectively.Results: Convinced by colonoscopy or surgery, there were 21 inflammation cases, 32 polyp cases, 24 tumor cases and 20 normal cases. The non-expert doctors diagnosed 49 out of 57 polyps in all of 37 cases, aroused 11 compared with using CTC only. The sensitivity was 91.67% for polyps whose diameter≤5mm, the result raised slightly. The non-expert doctors diagnosed 20 polyps whose diameter from 6 to 9 mm with CAD combining with CTC. The detection rate raised obviously and reached to 86.96%. For the polyps whose diameter≥10mm, the sensitivity of CAD reached to 81.82%.The non-expert readers diagnosed 19 tumor out or 22 in all, raising 2 compared with using CTC only, diagnosed 18 inflammation out or 21 in all, raising 3 compared with using CTC only, diagnosed 15 normal cases out or 17 in all, raising 2 compared with using CTC only.The expert doctors diagnosed 51 out of 57 polyps in all of 37 cases, aroused 3 compared with using CTC only. The sensitivity was 83.33% for polyps whose diameter≤5mm, the result had no statistical differences. The detection rate raised obviously and reached to 91.30% for polyps whose diameter from 6 to 9 mm. For the polyps whose diameter≥10mm, the result had no statistical differences.The non-expert readers diagnosed 21 tumor out or 22 in all, the same as using CTC only, diagnosed 20 inflammation out or 21 in all, raising 3 compared with using CTC only, diagnosed all 17 normal cases.It showed positive correlation between the diagnosing results of non-expert readers with/without CAD and the coefficient is 0.337, P<0.05. The diagnosing results of CAD combining with CTC showed high correlation with the results of endoscopy or surgery. The correlation coefficient is 0.703, P<0.01. It showed high correlation between the diagnosing results of non-expert readers with/without CAD and the coefficient is 0.743, P<0.01. The diagnosing results of CAD combining with CTC showed high correlation with the results of endoscopy or surgery. The correlation coefficient is 0.912, P<0.01.Conclusion: The detection rate raised by using CTC combing with CAD compared with using CTC only for both the expert and non-expert readers, especially for the non-expert doctors. It had advantages of using CTC combing with CAD in qualitative diagnosing compared with using CTC or CAD only, but still slightly lower than endoscopy. As a result, it was a nice non-invasive method to diagnose colon diseases by using CAD combining with CTC which was capable to reach the detection level of endoscopy to some extent and was promising to take the place of endoscopy in screening text for colon diseases.