ABSTRACT
The tumor-stroma ratio (TSR) has been repeatedly shown to be a prognostic factor for survival prediction of different cancer types. However, an objective and reliable determination of the tumor-stroma ratio remains challenging. We present an easily adaptable deep learning model for accurately segmenting tumor regions in hematoxylin and eosin (H&E)-stained whole slide images (WSIs) of colon cancer patients into five distinct classes (tumor, stroma, necrosis, mucus, and background). The tumor-stroma ratio can be determined in the presence of necrotic or mucinous areas. We employ a few-shot model, eventually aiming for the easy adaptability of our approach to related segmentation tasks or other primaries, and compare the results to a well-established state-of-the art approach (U-Net). Both models achieve similar results with an overall accuracy of 86.5% and 86.7%, respectively, indicating that the adaptability does not lead to a significant decrease in accuracy. Moreover, we comprehensively compare with TSR estimates of human observers and examine in detail discrepancies and inter-rater reliability. Adding a second survey for segmentation quality on top of a first survey for TSR estimation, we found that TSR estimations of human observers are not as reliable a ground truth as previously thought.
ABSTRACT
Purpose: Automatic outlining of different tissue types in digitized histological specimen provides a basis for follow-up analyses and can potentially guide subsequent medical decisions. The immense size of whole-slide-images (WSIs), however, poses a challenge in terms of computation time. In this regard, the analysis of nonoverlapping patches outperforms pixelwise segmentation approaches but still leaves room for optimization. Furthermore, the division into patches, regardless of the biological structures they contain, is a drawback due to the loss of local dependencies. Approach: We propose to subdivide the WSI into coherent regions prior to classification by grouping visually similar adjacent pixels into superpixels. Afterward, only a random subset of patches per superpixel is classified and patch labels are combined into a superpixel label. We propose a metric for identifying superpixels with an uncertain classification and evaluate two medical applications, namely tumor area and invasive margin estimation and tumor composition analysis. Results: The algorithm has been developed on 159 hand-annotated WSIs of colon resections and its performance is compared with an analysis without prior segmentation. The algorithm shows an average speed-up of 41% and an increase in accuracy from 93.8% to 95.7%. By assigning a rejection label to uncertain superpixels, we further increase the accuracy by 0.4%. While tumor area estimation shows high concordance to the annotated area, the analysis of tumor composition highlights limitations of our approach. Conclusion: By combining superpixel segmentation and patch classification, we designed a fast and accurate framework for whole-slide cartography that is AI-model agnostic and provides the basis for various medical endpoints.
ABSTRACT
Peritumoral budding and intratumoral budding (ITB) are important prognostic factors for colorectal cancer patients. Scientists worldwide have investigated the role of budding in tumor progression and its prognosis, but guidelines for reliably identifying tumor buds based on morphology are lacking. In this study, next-generation tissue microarray (ngTMA®) construction was used for tumor bud evaluation, and highly detailed rule-out annotation was used for tumor definition in pancytokeratin-stained tissue sections. Initially, tissues of 245 colon cancer patients were evaluated with high interobserver reliability, and a concordance of 96% was achieved. It was shown that high ITB scores were associated with poor distant metastasis-free survival (p = 0.006 with a cut-off of ≥10 buds). This cut-off was defined as the best maximum value from one of two/three ngTMA® cores (0.6 mm diameter). ITB in 30 cases of mucinous, medullary, and signet ring cell carcinoma was analyzed for the subsequent determination of differences in tumor bud analyses between those subtypes. In conclusion, blinded randomized punched cores in the tumor center can be useful for ITB detection. It can be assumed that this method is suitable for its adoption in clinical routines.
Subject(s)
Adenocarcinoma, Mucinous/pathology , Carcinoma, Medullary/pathology , Carcinoma, Signet Ring Cell/pathology , Colonic Neoplasms/pathology , Adenocarcinoma, Mucinous/metabolism , Biomarkers, Tumor/metabolism , Carcinoma, Medullary/metabolism , Carcinoma, Signet Ring Cell/metabolism , Colonic Neoplasms/metabolism , Disease Progression , Female , Gene Expression Regulation, Neoplastic , Humans , Male , Neoplasm Metastasis , Neoplasm Staging , Prognosis , Survival Analysis , Tissue Array AnalysisABSTRACT
BACKGROUND: A prototype system for computer-assisted colposcopic diagnosis (CAD) currently achieves a high level of accuracy of 80% (sensitivity 85%, specificity 75%) for the automatic assessment of colposcopic images. This pilot study investigated whether this type of CAD system is, in principle, capable of influencing the quality of the examiner's assessment. MATERIALS AND METHODS: In this observer study, 24 digitized colposcopic images from patients attending a dysplasia clinic were assessed by 90 participants. All participants had attended a colposcopy training workshop so that they acquired the same basic information and skills. RESULTS: Wide variation was seen among the non-experts, in contrast to the experts. An overall improvement in diagnostic accuracy was noted when the CAD system was used (non-experts: sensitivity 78%, specificity 70%; experts: sensitivity 74%, specificity 70%). CONCLUSION: The CAD system may serve as an aid in the further diagnosis of cervical intraepithelial neoplasia, and has the potential to improve the diagnostic process.
Subject(s)
Colposcopy/methods , Diagnosis, Computer-Assisted/methods , Uterine Cervical Dysplasia/diagnosis , Uterine Cervical Neoplasms/diagnosis , Colposcopy/education , Colposcopy/standards , Diagnosis, Computer-Assisted/standards , Female , Humans , Observer Variation , Pilot Projects , Sensitivity and Specificity , Uterine Cervical Neoplasms/pathology , Uterine Cervical Dysplasia/pathologyABSTRACT
PURPOSE: Diagnosis of cervical intraepithelial neoplasia (CIN) is currently based on the histological result of an aiming biopsy. This preliminary study investigated whether diagnostics for CIN can potentially be improved using semiautomatic colposcopic image analysis. METHODS: 198 women with unremarkable or abnormal smears underwent colposcopy examinations. 375 regions of interest (ROIs) were manually marked on digital screen shots of the streaming documentation, which we provided during our colposcopic examinations (39 normal findings, 41 CIN I, and 118 CIN II-III). These ROIs were classified into two groups (211 regions with normal findings and CIN I, and 164 regions with CIN II-III). We developed a prototypical computer-assisted diagnostic (CAD) device based on image-processing methods to automatically characterize the color, texture, and granulation of the ROIs. RESULTS: Using n-fold cross-validation, the CAD system achieved a maximum diagnostic accuracy of 80% (sensitivity 85% and specificity 75%) corresponding to a correct assignment of abnormal or unremarkable findings. CONCLUSIONS: The CAD system may be able to play a supportive role in the further diagnosis of CIN, potentially paving the way for new and enhanced developments in colposcopy-based diagnosis.
Subject(s)
Colposcopy/methods , Diagnosis, Computer-Assisted/methods , Uterine Cervical Dysplasia/diagnosis , Uterine Cervical Neoplasms/diagnosis , Adult , Aged , Cervix Uteri/pathology , Female , Humans , Image Processing, Computer-Assisted , Middle Aged , Reproducibility of Results , Sensitivity and Specificity , Vaginal SmearsABSTRACT
PURPOSE: To compare diagnostic performance and interobserver variability in a group of 36 examiners, with four different levels of experience. METHODS: Nine junior trainees, eight level I senior trainees, 11 level II senior gynecologists, and eight level III expert sonologists classified 105 ultrasound images of adnexal masses into three subgroups of ovarian lesions (malignancies, functional cysts, and dermoid cysts). RESULTS: The level III sonologists obtained the best diagnostic results together with the lowest interobserver variability (κ = 0.70, SD = 0.04). They achieved significantly better results in comparison with the junior trainees and also the senior trainees (κ = 0.51, SD = 0.12, p < 0.001; and κ = 0.51, SD = 0.09, p < 0.001). Differences between level III sonologists and the group of level II observers did not reach statistical significance (κ = 0.65, SD = 0.09, p = 0.70). There were no significant differences between senior and junior trainees (p = 1.0) and both groups achieved a significantly poorer diagnostic performance in comparison with the level II observers (p < 0.01 and p < 0.01). For all observers, the largest differences were seen for classifying malignancies, the best results for classifying functional cysts, and the poorest for evaluating dermoid cysts. CONCLUSIONS: Diagnostic performance of pattern recognition significantly improves with an increasing level of experience, emphasizing the importance of standardized ultrasound training programs with supervision by experts.
Subject(s)
Ovarian Diseases/pathology , Ovary/pathology , Pattern Recognition, Visual , Adolescent , Adult , Aged , Aged, 80 and over , Female , Humans , Middle Aged , Observer Variation , Organ Size , Ovarian Diseases/diagnostic imaging , Ovary/diagnostic imaging , Ultrasonography , Young AdultABSTRACT
OBJECTIVE: The objective of this study was to analyze the reliability of a landmark-independent method for determining the facial symmetry plane and degree of asymmetry based on three-dimensional data from the facial surface from two sets of recordings, one performed consecutively and one performed on different days. MATERIALS AND METHODS: We used an optical 3D-sensor to obtain the facial data of one male subject in two sets of ten measurements: the first taken consecutively and the second on different days. The symmetry plane and degree of asymmetry were calculated for each of the resulting twenty sets of data. One set of data was analyzed ten times for control purposes. The calculation of the mean deviation angle between the symmetry planes served as a measure of the reproducibility of these results. RESULTS: Although the mean angular deviations of the computed symmetry planes, 0.134 degrees (for ten consecutively captured images) and 0.177 degrees (for the ten images captured on different days), were each significantly higher than the mean angular deviation (0.028 degrees) calculated from ten analyses of a single image, they can still be regarded as very small. There were no significant differences in the degree of asymmetry among the three measurement sets. The standard deviations revealed low values. CONCLUSIONS: This method can be used to compute with high reliability the symmetry planes and degree of asymmetry of facial 3D-data. The color-coded visualization of asymmetrical facial regions makes it possible for this analytical procedure to capture the asymmetries of facial soft tissue with substantially greater precision than 2-dimensional en face images.
Subject(s)
Facial Asymmetry/diagnosis , Image Processing, Computer-Assisted/instrumentation , Imaging, Three-Dimensional/instrumentation , Orthodontics, Corrective , Photogrammetry/instrumentation , Software , Cephalometry/instrumentation , Computer Graphics , Equipment Design , Facial Asymmetry/therapy , Humans , Male , Sensitivity and SpecificityABSTRACT
PURPOSE: This study reports on the intraoperative use of noncontact, nonionizing, optical 3-dimensional (3D) exophthalmometry during the repositioning of dislocated globes as a result of trauma. PATIENTS AND METHODS: Ten patients (4 female, 6 male, 41.4+/-15.2 years) with a relative enophthalmos of the globe as a result of zygomatic fractures were included in the study. Preoperatively, en- and exophthalmometry data were assessed from axial CT slices and optical 3D imaging. 3D data were analyzed twice for the assessment of measurement errors. Intraoperatively, optical en- and exophthalmometry was carried out to control the globe position. Surgery was considered successful when the relative en- or exophthalmos no longer exceeded 2 mm. Optical 3D en- and exophthalmometry data were reassessed 5 days and 3 months after surgery. RESULTS: Method error was 0.184 mm for optical 3D en- and exophthalmometry. The preoperatively assessed en- and exophthalmometry data determined from axial CT scans and from optical 3D images did not differ significantly statistically (P=.538). When the preoperative en- and exophthalmometry data were compared to the values assessed at the end of surgery, a significant improvement in globe position was found (P=.005). Although a relative en- or exophthalmos of 2 mm was not exceeded in any of the patients 3 months after surgery, en- and exophthalmometry data differed significantly statistically from the data assessed at the end of the operation (P=.005). CONCLUSIONS: Intraoperative optical en- and exophthalmometry is an effective means to support the surgeon in objectively optimizing the globe position with small measurement errors.
Subject(s)
Diagnostic Techniques, Ophthalmological , Exophthalmos/diagnosis , Adult , Enophthalmos/diagnosis , Enophthalmos/etiology , Exophthalmos/etiology , Female , Humans , Imaging, Three-Dimensional/methods , Intraoperative Care , Male , Orbital Fractures/complications , Reproducibility of Results , Zygomatic Fractures/complicationsABSTRACT
BACKGROUND: Several severe complications have been described with blow-in fractures. Therefore, immediate surgical treatment of these fractures has been recommended. To date, there is only minimal knowledge on long-term complications of blow-in fractures that have remained untreated. The present case report describes a late complication of an untreated blow-in fracture of the orbital floor. CASE: A 37-year-old male was involved in a car accident 16 years before. At that time, a non-dislocated midfacial fracture was diagnosed and remained untreated because of the lack of clinical symptoms. Four months before surgery an exophthalmos of the left globe began to develop. CT examination revealed a consolidated blow-in fracture of the left orbital floor and an opaque mass around the dislocated bony fragments. By an infraorbital approach the bony fragments and the surrounding mass were removed. Histological examination of the removed material revealed a cystic structure lined with respiratory epithelium. Therefore, the diagnosis 'post-traumatic mucocele in the orbit caused by dislocated respiratory epithelium from the maxillary sinus' was made. CONCLUSION: Even if blow-in fractures do not cause complications immediately after trauma, late complications like mucoceles can occur after several symptom-free years. Therefore, early reconstruction should be intended even in asymptomatic cases of blow-in fractures with minimal displacement of the bony fragments.
Subject(s)
Mucocele/etiology , Orbital Fractures/complications , Accidents, Traffic , Adult , Exophthalmos/etiology , Exophthalmos/surgery , Humans , Male , Mucocele/surgery , Orbital Fractures/surgeryABSTRACT
AIM: It is the aim of the present study to introduce non-contact, non-invasive optical 3D imaging to relative exophthalmometry and to compare the resulting data to exophthalmometry values assessed by the Hertel instrument and computed tomography. PATIENTS AND METHODS: 20 patients (3 female, 17 male, 44.4+/-16.6 years) without orbital pathology, who were examined by computed tomography of head and neck for the exclusion of different diseases, and seven patients (1 female, 6 male, 40.1+/-14.4 years), who received routine orbital computed tomography because of zygomatic fractures, were included in the study. Optical 3D images of the facial surface were assessed and Hertel exophthalmometry was carried out to determine the relative globe position. In patients with zygomatic fractures the assessment of optical 3D images and Hertel values was repeated 5 days after surgery. RESULTS: For patients without orbital pathology relative exophthalmometry data were 1.4+/-1.1 mm for the Hertel instrument, 0.9+/-1.0 mm for computed tomography and 0.5+/-0.5 mm for optical 3D imaging. The values for Hertel exophthalmometry and computed tomography did not differ statistically significantly (p(Herteldifferencepreop/CTdifferencepreop)=0.284), while there was a significant difference between Hertel exophthalmometry and optical 3D imaging (p(Herteldifferencepreop/opticaldifferencepreop)=0.008). In the cases of zygomatic fractures, Hertel exophthalmometry revealed less pronounced relative differences in globe position than CT and optical 3D imaging data (Hertel 0.7+/-1.1 mm, CT 1.9+/-1.0 mm, optical 3D imaging 1.9+/-1.0 mm). Postoperatively, relative Hertel exophthalmometry showed an increased value revealing a more pronounced enophthalmos (1.7+/-1.0 mm), while the corresponding value of the optical 3D images decreased as a sign for normalization of the globe position (1.1+/-0.7 mm). CONCLUSION: Because of its reliance on the lateral orbital rims Hertel exophthalmometry can lead to an under- or overestimation of enophthalmos, when soft tissue oedema or a dislocation of the orbital rim are present. The combination of computed tomography as baseline measurement and optical 3D imaging for the follow-up examinations reveal more realistic data in cases of zygomatic fractures. Therefore, they should be preferred to the determination of Hertel values especially in more complex cases.
Subject(s)
Enophthalmos/diagnosis , Exophthalmos/diagnosis , Imaging, Three-Dimensional/methods , Tomography, X-Ray Computed , Zygomatic Fractures/complications , Adult , Diagnostic Techniques, Ophthalmological/instrumentation , Eye/diagnostic imaging , Eye/pathology , Female , Follow-Up Studies , Humans , Image Processing, Computer-Assisted , Male , Optics and Photonics/instrumentation , Orbit/diagnostic imaging , Orbit/pathology , Photography/instrumentation , Statistics, Nonparametric , Zygomatic Fractures/surgeryABSTRACT
The purpose of this study was to validate the assessment of visible volume changes of the facial soft tissue with an optical three-dimensional sensor and to introduce new parameters for the evaluation of the soft-tissue shape achieved from three-dimensional data of selected cases of midfacial distraction. Images of a truncated cone of known volume were assessed repeatedly with an optical three-dimensional sensor based on phase-measuring triangulation to calculate the volume. Two cubic centimeters of anesthetic solution was injected into the right malar region of 10 volunteers who gave their informed consent. Three-dimensional images were assessed before and immediately after the injections for the assessment of the visible volume change. In five patients who underwent midfacial distraction after a high quadrangular Le Fort I osteotomy, three-dimensional scans were acquired before and 6 and 24 months after the operation. The visible soft-tissue volume change in the malar-midfacial area and the mean distance of the accommodation vector that transformed the preoperative into the postoperative surface were calculated. The volume of the truncated cone was 235.26 +/- 1.01 cc, revealing a measurement uncertainty of 0.4 percent. The injections of anesthetic solution into the malar area resulted in an average visible volume change of 2.06 +/- 0.06 cc. The measurement uncertainty was 3 percent. In the five patients, the average distance of maxillary advancement was 6.7 +/- 2.3 mm after 6 months and 5.4 +/- 3.0 mm after 2 years. It was accompanied by a mean visible volume increase of 8.92 +/- 5.95 cc on the right side and 9.54 +/- 4.39 cc on the left side after 6 months and 3.54 +/- 3.70 cc and 4.80 +/- 3.47 cc, respectively, after 2 years. The mean distance of the accommodation vector was 4.41 +/- 1.94 mm on the right side and 4.74 +/- 1.32 mm on the left side after 6 months and 1.62 +/- 1.96 mm and 2.16 +/- 1.52 mm, respectively, after 2 years. The assessment of visible volume changes by optical three-dimensional images can be carried out with considerable accuracy. The determination of volume changes and accompanying accommodation vectors completes the cephalometric analysis during the follow-up of patients undergoing midfacial distraction. The new parameters will help to assess normative soft-tissue data on the basis of three-dimensional imaging with a view to an improved three-dimensional prediction of the operative outcome of orthognathic surgery.
Subject(s)
Cephalometry , Face/anatomy & histology , Facial Bones/abnormalities , Image Processing, Computer-Assisted , Imaging, Three-Dimensional , Maxilla/abnormalities , Maxilla/surgery , Osteogenesis, Distraction , Photography/instrumentation , Adolescent , Adult , Child , Female , Humans , Male , Osteotomy, Le Fort/rehabilitation , Photography/methodsABSTRACT
OBJECTIVE: To evaluate and compare the effects of early primary closure of the hard palate on the anterior and posterior width of the maxillary arch in children with bilateral (BCLP) and unilateral (UCLP) cleft lip and palate during the first 4 years of life. DESIGN: A retrospective, mixed-longitudinal study. SETTING: Cleft Palate Center of the University of Erlangen-Nuremberg. SUBJECTS AND METHODS: The present investigation analyzes longitudinally 42 children with UCLP and 8 children with BCLP between 1996 and 2000 with early simultaneous primary closure of lip and hard palate (4 to 5 months). Palatal arch width was measured on dental casts with a computer-controlled three-dimensional digitizing system, and their growth velocities were calculated from consecutive periods (mean follow-up 39 months). Differences in growth velocities were compared with those of 25 children with UCLP and 15 children with BCLP with delayed closure of hard palate (12 to 14 months). RESULTS AND CONCLUSIONS: There was no significant difference in terms of anterior and posterior maxillary width between early and delayed closure of hard palate within the first 4 years of life.