What Student Need to Know about Overview of Research Designs Applied In Radio-Diagnosis.

 What Student Need to Know about Overview of Research Designs Applied In Radio-Diagnosis.

Research in the field of radiology and imaging sciences has generated critical technologies, including magnetic resonance imaging (MRI), computed tomography (CT), and nuclear medicine as well as honed modern techniques for using equipment to identify abnormalities and disease to be treated. Today’s research is seeking discoveries to contribute to advances in the treatment of diseases such as cancer and Alzheimer’s disease.

Performing methodologically rigorous scientific research is not a trivial task. The optimal research study will be directed at an important, precisely defined clinical question, with a specified target population matched by the subject selection. The most efficient study design will be used and the sample size will be sufficient to limit type II error to an acceptable level. Further, bias will be avoided, and the results will be reliable, internally valid, and generalizable to the target population and possibly beyond. Success at such demanding research endeavors is certainly within the reach of radiologists and radiology researchers.

Additional potential biases in diagnostic test evaluation include spectrum bias, in which only patients with overt disease are used in assessment of a diagnostic test. Not including subtle or indeterminate cases can also lead to overestimation of disease accuracy. Prospective data collection is generally less subject to bias than retrospective collection and is therefore preferred when designing a study. However, retrospective data collection may be preferred in a few circumstances, such as when prospective data collection would remove the ability to blind the observers and would therefore potentially introduce greater bias.

In general, studies with bias tend to report more encouraging results than those without bias. In addition, preliminary studies of a diagnostic technology, performed with small sample size and vulnerable to bias, often will be highly optimistic about the capabilities of that technology. Subsequent reports may present a more realistic appraisal.

The Research Question

The first step in any research endeavor is to frame an appropriate research question. This question must be important (or it is not worth our efforts), but it also must be precise. As an example, we can start with a common and vexing clinical problem that has been the cause of considerable interest in the radiology literature, “Which test is better in patients with possible appendicitis, CT or sonography?” This question is certainly important and clinically relevant, but as framed above it cannot be answered.

The question must be defined more precisely with respect to the type of patients in whom the question is being raised, the target population, and what is actually being asked. The imaging accuracy and usefulness of sonography and CT will likely vary on the basis of a number of patient-specific variables. Are the patients we are interested in adults or children? Are they thin or fat? Are they cooperative or un-cooperative? Are they men or women? Disease-specific factors may also affect the imaging. Has the patient been symptomatic for a few hours and we suspect simple unperforated appendicitis, or has the patient been symptomatic for 4 days and appears septic, leading us to suspect an abscess? These factors also might affect the performance of sonography and CT.

Study Design

Having determined the question to be answered, the next issue is the research methodology itself. To produce evidence that will appropriately drive decision making, experimental design is of critical importance and will be the focus of much of this article series. The goal of study design is to achieve the most with the least (i.e., to achieve efficiency). Fortunately, we have the experience of clinical epidemiologists and biostatisticians with decades of experience from which to draw to determine the most efficient way of designing studies and the most appropriate way to productively critique research. Prospective comparisons of diagnostic test results with a well-defined reference test and randomized double-blinded clinical trials are the study designs that provide the best information to guide clinical practice


The opposite of random error is systematic error that is introduced through inadequacy in the study design, subject selection, or analysis. Statistics are for the most part unable to compensate for systematic error. Avoidance of such systematic error, or bias, is one of the major challenges of research design. Unfortunately, many of the apparently simple research designs that are common in the radiology literature succumb to bias.

#British #BioMedicine #Institute #Evidence #Skill #eLearning #Platform #BritishCancerInstitute #AIIMSDelhi #AIIMSRishikesh #bbminstitute #bbmclinicaltrials #britishbiomedicine #BritishHealthcareImagingInstitute #BHII #BCI #bjpmr #bjbmr #BSGM #BSHA #BSPM #BSRD #BSDM #BBMI #BSCR #BSMD #BBMCT #BBM #PHM #CTPRA #CCP #MDRA #BSCCP #BBMI #NanoDegree #BritishSchoolOfClinicalResearch #BritishSchoolOfClinicalChildPsychology #BritishSchoolOfMedicalDevice #BritishSchoolOfYogaHealth #BritishBioMedicineClinicalTrials #BritishYogaHealth® #BritishChildPsychology #BritishBioMolecule #ExclusiveAIIMSHospital #Dermatologytrials #CardiovascularTrials #DiabetesTrials #OncologyTrials #HematologyTrials #PediatricTrials #CovidTrials #NeuroScienceTrials #GyneacologyTrials #GastroenterologyTrials #RareDiseaseTrials #AutoImmuneTrials #InfectiousDiseaseTrials #EndocrineTrials #OpthalmologyTrials #NephrologyTrials #ConductClinicalTrials #directorbbmclinicaltrialscom #checkyourplagiarism #NewApprovedDrug  #YogaTeacherTrainingProgram #YTTP #BYH #100YTTP #200YTTP #300YTTP #500YTTP #YogaProfessor #BritishSchoolOfPainManagement #BritishSchoolOfGenomicMedicine #BritishSchoolOfHospitalAdministration #BritishSchoolOfRadioDiagnosis #BritishSchoolOfPublicHealthManagement #BritishSchoolOfDiabetesManagement

As an example, one could imagine a study designed to compare CT and MR imaging for detection of liver metastases in patients with known adenocarcinoma of another organ. To identify patients for such a study, one might review all the patients who underwent both tests, and using some external gold standard, make a comparison. However, would this study design be free of bias? Likely, there would be significant bias in the selection of the subjects. For example, if at a given center CT is generally used as the initial imaging modality for the evaluation of possible liver metastases, then the patients who undergo both imaging studies would be the ones in whom the initial CT was equivocal. The comparison would not be CT versus MR imaging, but rather, CT versus MR imaging in patients in whom the CT was equivocal. Of course, the results of such a study would underestimate the accuracy of CT, because only those cases that are difficult to diagnose with CT were included. This is a simple but unfortunately common example of selection bias in recruiting patients for a study. Selection bias occurs when the subjects studied are not representative of the target population. In the previous example, the target population is all patients with known adenocarcinoma of another organ. However, the study group is only those patients with known adenocarcinoma who underwent both CT and MR imaging. To avoid this bias, subject selection should be based on clinical criteria (i.e., all subjects with a new diagnosis of adenocarcinoma) rather than availability of imaging studies

Data Analysis

Research is conducted on samples. We measure outcome or accuracy on a relatively small number of subjects. Yet the intent of research is (eventually) to influence clinical care. To achieve this, the research results must be valid on subjects other than those included in the study. Statistics is the science that allows us to make inferences about populations from measurements made on samples. A vast array of tools is available to the biostatistician to enable such inference. These tools must be familiar to the research radiologist and will be discussed in future modules. In this discussion I will limit myself to introduction of the concepts of validity and reliability.

Validity can be divided into internal validity and external validity, which is also known as generalizability. Internal validity refers to the extent to which the results and conclusions of a study actually relate to true events in the sample under study. Some of the biases and study design considerations described previously relate to validity. For example, an observer who is aware of the results of the reference test might unintentionally overestimate the accuracy of the diagnostic test under study. Thus, the recorded results might not be an internally valid representation of the actual sample. The method of data analysis and the statistical tests used are also critical to the internal validity of the study, because use of inappropriate analysis can lead to false conclusions.

Similarly, the external validity of a study is dependent on both the research design and the analytic methods. The extent to which the sample selected truly reflects the target population is a strong determinate of the generalizability of a study. Also, the use of appropriate statistics allows determination of what inferences can be drawn about the target population on the basis of the sample data.

Leave a Reply

Subscribe to
our Newsletter

***We Promise, no spam!

British Healthcare & Imaging Institute Was Conceived As A Premier Institute Catering To The Ever Increasing And Demanding Task Of Providing World Class Medical Diagnostic Solutions Under One Roof.

We’re Available

Monday : 08.00 - 6.00
Tuesday : 08.00 - 6.00
Wednesday : 08.00 - 6.00
Thursday : 08.00 - 6.00
Friday : 08.00 - 6.00
Saturday : 10.00 - 05.00
Sunday : 10.00 - 02.00