Study On Medical Imaging Cases Library’s Building And Its Clinical Value In The Diagnosis Of Mediastinal Space Occupying Lesions

Part I The building of medical imaging cases library based on PACSObjectiveStudying the medical information keywords of certain standard term related to medical imaging diagnosis and building the imaging cases library of clinical diseases based on PACS aim at quickly obtaining the required image datas of the past for Radiologists in the process of diseaes diagnosis, so as to improve the accuracy of disease diagnosis.Materials and methods1.Clinical dataFrom May 2010 to Dec.2015242 cases were selected among the patients diagnosed as mediastinal space occupying lesions by pathology. Researching certain keywords related to medical imaging diagnosis and building the keywords library in the PACS on the basis of the imaging information of all kinds of diseases. Building retrieval system based on the keywords by associating hospital EMR system.2.Equipment and methods2.1 The image informatics keywords set of imaging cases libraryThe heart of the retrieval system is to have complete keywords. By studying the imaging diagnosis of disease and clinical research, combined with the hospital EMR, we have extracted the following keywords set. Including patient basic information, clinical diagnose, imaging information, clinical inspection information, treatment method and postoperative situation and so on.2.2The management of the imaging cases library(1)Case registration:The patient physical information, clinical diagnose and inspection information have been a structured data unit in hospital information system. But the imaging information is basically a kind of free description form, not a standard data unit. Therefore, case registration is mainly for input of imaging information.(2)Case management:For each registered case, professional radiologist is required to check the new registered case, make some correction, modification and supplement, and also delete some cases that do not conform to the standards.2.3The search methods of imaging cases library(1)Based on statistical information:a.patient’s name, gender, the name of disease, etc; b.encoding/code or their combinations, such as diagnostic code; c.quantitative values or their scope, such as age/age group, date or time, etc.(2)Based on the imaging information:a.imaging technology and method; b.anatomical location; c.imaging characteristic of lesion; d.imaging name of disease, etc.ResultsTaking mediastinal teratoma for example to illustrate. A 26-year-old female patient was found mediastinum space occupying focus after CT examination. The lesion was located in the anterior mediastinum with irregular shape, fuzzy boundary, cyst-solidary nature and its surrounding tissues were compressed and displaced. The density of lesions was uneven, it contained calcium and fat composition. After contrast-enhanced scan the lesion presented uneven enhancement. The doctor was doubtful whether the disease was teratoma. Now if you want to make a further definitive diagnosis for the disease according to existing CT imaging findings. For some radiologists, especially the one who lacks of clinical experience, without the result of pathological examination, the most important way is to refer to the previous imaging cases of clinically diagnosed. Using the imaging cases library established above, the detailed processes and results of searching the related cases are following. (1)If you option "teratoma" in "imaging diagnosis and impression" of the imaging informations, the query result of related cases is 18; (2)If you option "anterior mediastinum" in "definite location" on the basis of (1), the cases reduce to 15. (3)If you again select "calcification and fat composition" in "other characteristics", the cases become 8. (4)If you choose gender (female) and age (20 to 30 years) in "patient basic information", the cases once again reduce to 3. Then you can analyze these 3 cases by comparison, one of the patients diagnosed teratoma is extremely similar to the case. Finally, by pathologic result the disease is confirmed into teratoma.ConclusionBuilding the medical imaging cases library of mediastinal space occupying lesions based on PACS will help doctors to diagnosis disease, improve the preoperative diagnostic accuracy.Part II Clinical characteristics, imaging performances and pathologic features of mediastinal space occupying lesionsObjectiveRetrospectively analyzed the clinical and imaging data about 242 cases of pathologically confirmed mediastinal space occupying lesions. Understanding all kinds of tumors and different pathological subtypes on their clinical features and image manifestations according to WHO2015 version of the new classification of mediastinal space occupying lesions to provide necessary reference for their epidemiological studies,image diagnosis and differential diagnosis.Materials and methods1.Clinical dataCollecting 242 cases confirmed mediastinal space occupying lesions by surgery, pathology biopsy or bronchoscopy from 2010/5 to 2015/12 in our hospital, male 127 cases, female 115, age from 2 to 79 years old, the average is 41.9. Clinical manifestations contain chest pain, chest tightness, shortness of breath, difficulty breathing and swallowing, facial and upper limb swelling and so on, parts have no obvious clinical symptoms.2.Equipment and methodsUsing Siemens SOMATOM Definition dual-source CT, GE Lightspeed 16 row CT and Philips offerings iCT, all patients were performed CT examination, plain scan 39 cases, enhance scan 203. Contract materials were medium ioversol injecting in elbow intravenous by high pressure syringe. All scan datas were sent to the PACS workstation for MPR. Finally 2 radiologists diagnosed together. The main imaging analysis of lesions are as follows:location, shape, size, density, boundary, edge, necrosis and cystic, calcification, fat content, strengthening characteristics, relationship to the adjacent, the clear degree of mediastinal fat layer and other abnormal performance in mediastinum or bilateral pleural cavity.3.Pathological examination method157 cases underwent pathological examination after operation,76 of CT guided mediastinal fine needle biopsy examination,9 bronchoscopy. The specimen fixed with 4% formaldehyde,8% hydrochloric acid and formic acid for decalcification, sliced paraffin embedding and routine HE staining.Then observed the histomorphology, some underwent immunohistochemical staining.4.Statistical analysisAnalysis was carried on by using the SPSS 21.0, significant level is 5%, P-value less than 5% was defined as had statistical significant. Compare the gender, age distribution, the benign and malignant incidence in primary and secondary mediastinal space occupying lesions and location with chi-square test. Compare the imaging characteristics of benign and malignant lesions with multiple independent samples nonparametric test (Wilcoxon rank and inspection).Results1.The sex and age of mediastinal space occupying lesionsIn 242 cases, male were 127 cases, female 115, age was from 2-79 years old, average 41.9, median age 43. Group 0～10 years had 18 cases,11～20 were 21,21-30 were 44,31-40 were 37,41 to 50 years 45,51 to 60 years 42,61-70 years 33,71-80 years 10.2.The classifications and pathology types of mediastinal space occupying lesionsIn 242 cases, benign lesions were 143 cases(59.1%) with 130 primary and 13 secondary, including 60 cases(24.8%) from thymus, of which 45 were thymoma(type A 8 cases,11 type B1,9 type B2,4 type B3, type AB 8 and mixed type 5),8 thymic hyperplasia,4 thymic lipoma and 3 thymic cyst. Germ-cell tumors were 15 cases (6.2%), all were mature teratoma; Lymphatic tumors were 3 cases(1.2%), all were Castleman diseases. Neurogenic were 24 cases(9.9%), of which benign schwannoma were 19, ganglion cells neuroma 4, nerve fibroma 1. Thyroid and parathyroid were 9 cases(3.7%), including 7 cases of nodular goiters, thyroid cyst and parathyroid adenoma 1 case respectively.6 cases(2.5%) from mesenchymal tissue, lipoma, cystic lymphangioma, fake myoid hemangioendothelioma, lymphatic hemangioma, ligament fibromatosis and benign isolation fibroma 1 case respectively. Trachea and esophagus tumors were 21 cases(8.7%), including bronchial cyst 16 cases, esophageal leiomyoma 2, esophageal cyst, esophageal tuberculosis and abscess 1 case respectively. Others were 5 cases(2.1%), including mediastinal hematoma 2 cases, pulmonary abscess, pulmonary sclerosing hemangioma and superior vena cava tumor 1 case respectively.Malignant lesions were 99 cases(40.9%) with 75 primary and 24 secondary, including thymic carcinoma 26 cases(10.7%)(12 squamous carcinoma,5 small cell neuroendocrine carcinoma,3 undifferentiated carcinoma,2 carcinoid tumor, adenocarcinoma, pleomorphic carcinoma, small cell carcinoma and lymphatic epithelioma 1 case respectively). Germ-cell tumors were 13 cases(5.4%), of which malignant teratoma were 3 cases, non-seminoma 5 cases(embryo sinus tumor 3 cases, embryonal carcinoma 2 cases), mixed type 4 cases and 1 seminoma. Lymphoma were 33 cases(13.6%), including non-hodgkin’s lymphoma 29 cases(B cell lymphoma 17 cases and T cell lymphoma 12, Hodgkin’s lymphoma 4 cases(2 nodular sclerosis type, lymphocytic type and no parting each 1 case). Neurogenic tumors were 3 cases(1.2%), of which 2 were malignant schwannoma and 1 neuroblastoma. Thyroid carcinoma were 2 cases(0.8%). Mesenchymal tissue source were 3 cases(1.2%), including myofibroblastic sarcoma, epithelioid angiosarcoma and malignant solitary fibrous tumor 1 case respectively. Esophageal cancer were 1 case(0.4%). Others were 18 cases(7.4%), including acute leukaemia mediastinal infiltration, mediastinal type lung cancer and metastases 6 cases respectively.3.The image manifestations of mediastinal space occupying lesions3.1 Location of lesionsIn anterior mediastinum, there were 131 cases(53.7%),77 benign and 54 malignant. The benign tumors included thymoma 43 cases, mature teratoma 12, thymic hyperplasia 7, thymus lipoma 3, bronchial cyst 3, thymic cyst 2, nodular goiter, fake myoid hemangioendothelioma, lymphangioma, superior vena cava tumor, chronic haematoma, lipoma and benign solitary fibrous tumor 1 case respectively. Malignant tumors included lymphoma 17 cases, thymic carcinoma 17, malignant germ-cell tumors 11, acute leukemia mediastinal infiltration 5, thyroid carcinoma 2, mediastinal type lung cancer and malignant solitary fibrous tumor each 1 case.In middle mediastinum, there were 34 cases(14.5%), each benign and malignant 17. The benigns included bronchial cyst 4 cases, Castleman diseases 3, nodular goiter 3, parathyroid adenoma and thyroid cyst, mature teratoma, esophageal abscess, thymus lipoma, lung abscess and pulmonary sclerosing hemangioma 1 case respectively. Malignants included metastatic tumor 6 cases, lymphoma 4, mediastinal type lung cancer 2, thymic carcinoma, malignant schwannoma, esophageal cancer, epithelioid angiosarcoma and myofibroblastic sarcoma 1 case respectively.In posterior mediastinum, there were 44 cases(18.2%),39 benign and 5 malignant. Benigns included schwannoma 19 cases, bronchus cyst 8, ganglion cells neuroma 4, esophageal leiomyoma 2, ligament fibromatosis, neurofibroma, mature teratoma, esophageal tuberculosis, lymphatic hemangioma and esophageal cyst 1 case respectively. Malignants had mediastinal type lung cancer 3 cases, neuroblastoma and malignant schwannoma 1 case respectively.There were 33 cases(13.6%) across mediastinum,10 benign and 23 malignant. Benigns included nodular goiter 3 cases, thymoma 2, thymic hyperplasia, chronic haematoma, mature teratoma, thymic cyst and bronchial cyst 1 case respectively. Malignants had 12 cases of lymphoma,8 thymic carcinoma,2 malignant germ-cell tumor and 1 acute leukemia mediastinal infiltration.3.2Image manifestationsFor benign lesions, the appearances were round like(86 cases), irregular(53) and triangle(4); the boundary were clear in 116 cases, edge neat 90, and uniform density 62.61 cases with cyst degeneration and necrosis,36 calcification and 16 fat.60 lesions with clear adjacent mediastinal fat layer,63 turbidity and 20 disappeared. After enhanced scanning,23 cases were abvious enhancement,33 moderate,52 mild and 12 cases without reinforcement (22 cases were plain scan).17 cases with pericardial or pleural effusion,4 with enlarged mediastinal lymph nodes.For malignant lesions, the appearances were round like(21 cases), irregular(77) and multiple nodular (1); the boundary were unclear in 88 cases, edge unneat 77, and uneven density 87.87 cases with cyst degeneration and necrosis,20 calcification and 4 fat.14 lesions with clear adjacent mediastinal fat layer,34 turbidity and 51 disappeared. After enhanced scanning,10 cases were abvious enhancement,40 moderate,33 mild and 15 cases were plain scan.49 cases with pericardial or pleural effusion,28 with enlarged mediastinal lymph nodes.4.Statistical analysis resultsThere were no statistical significance in gender among all kinds of tumors. There were statistical significance in age distribution or location among some common tumors(each P<0.001). There were statistical significance in the benign/malignant incidence rate between primary and secondary mediastinal space occupying lesions(P<0.001). There were statistical significance between benign and malignant in shape(P<0.001), boundary(P<0.001), edge(P<0.001), density(P<0.001), cyst degeneration and necrosis(P<0.001), fat(P=0.048), mediastinal fat layer(P<0.001), and paratumorous lesion(P<0.001). There were no statistical significance in calcification(P=0.368) and enhancement(P=0.817). There were statistical significance among all kinds of benign/malignant in cyst degeneration and necrosiss (P<0.001,0.031), calcification and fat(all P<0.001).Conclusion1.In mediastinal space occupying lesions, primary is more than secondary, benign than malignant. The primary lesions mainly come from thymus(mainly thymoma and thymic carcinoma), germ-cell tumor, lymphoma and neurogenic tumor. The secondary tumors mainly from thyroid, metastatic and mediastinal type lung cancer.2.Mediastinal space occupying lesions are more common in the younger adult patients, the gender ratio has no obvious difference. But the thymic tumor occurs in patient more than 40 years old, germ-cell tumor and lymphoma occur in patient under 40. Thymic carcinoma, malignant germ-cell tumor, extra-medullary infiltration of acute leukemia and metastases are more common in men.3.0n the basis of anatomical structure, the incidence in anterior mediastinum is higher than others. Thymic tumor and germ-cell tumor are more common in anterior mediastinum. Lymphoma occurs more in anterior mediastinum and anterior-middle mediastinum. Malignant mediastinum tumors ofen span multiple parts.4.The benigns are almost round-like shape with clear boundary, neat edge,while the malignants are always irregular shape with unclear boundary and unneat edge.5.Mediastinal space occupying lesions are often with uneven density, but the malignant are more obvious, and more necrosis and cystic lesions than benign. The fat is more common in benign tumors.6.On the change of mediastinal fat layer gap, malignant are more obvious than benign, and mediastinal malignant tumors are more ofen with pleural or(and) pericardial effusion and mediastinal lymph node enlargement.7.The necrosis and cystic lesions are often seen in thymoma, teratoma and neurogenic of benign tumors, of malignant tumors most found. Calcifcation is more seen in thymoma and teratoma of benigns, while in thymic carcinoma of malignants. Fat of benigns is often found in teratoma, rarely seen in malignant tumors.8.Patients with thymoma ofen accompanied with myasthenia gravis, while not the thymic carcinoma. The germ-cell tumor patients often accompany with laboratory indexes AFP and(or) P-HCG rise, especially in malignant tumors.9.Mesenchymal tissue tumors are rare in mediastinal, which lack of characteristic imaging findings and preoperative diagnosis are difficult.Part III The clinical value of medical imaging cases library on improving diagnosis of mediastinal space occupying lesionsObjectiveThe purpose of this part is mainly to use the medical imaging cases library discussed in part I to improve the preoperative diagnosis level of mediastinal space occupying lesions.Materials and methods1.Clinical dataCollecting 50 cases confirmed mediastinal space occupying lesions by pathology from 2015/09 to 2015/12 in our hospital, male 22 cases, female 28, age from 4 to 76 years old, the average is 40. Clinical manifestations contain chest pain, chest tightness, cough and shortness of breath and so on, some have no obvious clinical symptoms.2.Equipment and methodsCT examination as part II. All image datas were blindly reviewed and assessed by two imaging professional doctors with qualification, and obtained the diagnosis. The author then further analyze the mediastinal lesions of the 50 patients respectively for preoperative diagnosis using the medical imaging cases library(specific query methods as part Ⅰ), and obtained the diagnosis. The author called the latter(using the medical imaging cases library for preoperative diagnosis analysis) as experimental group, and the former(blindly review) as control group. Finally comparing the two sets of results with postoperative pathological results respectively, and contrasting preoperative diagnostic accuracy of the two groups. The preoperative diagnosis results according with the postoperative pathological results are referred to as positive, otherwise negative.3.Pathological examination methodPathological examination method as part II.4.Statistical analysisStatistical analysis as part II.Results1.The classification of mediastinal space occupying lesionsIn 50 cases, benign space occupying lesions had 29 cases(58%), including 11 thymoma,10 neurogenic tumor,3 nodular goiter,2 thymic hyperplasia,2 bronchial cyst and 1 Castleman Disease. Malignant had 21cases(42%), including 8 lymphoma, 6 thymic carcinoma,6 germ-cell tumor and 1 myofibrosarcoma.2.The preoperative diagnostic condition of the two methods for mediastinal tumor and tumor-like lesionsThere were 37 positive cases of preoperative diagnosis in the control group and negative were 13, the accuracy rate was 74%. while in the experimental group, the postive cases were 44 and negative 6, the accuracy rate was 88%. The negative cases of preoperative diagnosis in control group included 4 thymic carcinoma(3 were misdiagnosed as lymphoma,1 Castleman Disease),3 malignant germ-cell tumor(all were misdiagnosed as thymoma),3 thymoma(2 were misdiagnosed as teratoma,1 lymphoma),2 lymphoma(each were misdiagnosed as thymic carcinoma) and 1 myofibrosarcoma(misdiagnosed as teratoma). The negative cases in experimental group included 2 lymphoma(1 was middiagnosed as malignant germ-cell tumor and 1 thymoma),1 thymic carcinoma(misdiagnosed as lymphoma),1 thymoma(misdiagnosed as teratoma),1 myofibrosarcoma(misdiagnosed as teratoma), 1 bronchial cyst(misdiagnosed as pericardial cyst).3.Statistical analysis resultThere were no statistical significance in the preoperative diagnostic accuracy between the experimental group and control group(P=0.076).ConclusionIn comparision of the preoperative diagnostic accuracy for mediastinal space occupying lesions, although there are no statistical significance between using the medical imaging cases library and without, for the preoperative diagnostic accuracy, using the system is higher in our study. If enlarging the sample size, we believe that the preoperative diagnosis level of mediastinal space occupying lesions will be improved by using this system, and these two methods will have statistical significance.