Missing lesions on chest radiographs is frequent and the largest source of medicolegal issues. In this chapter, we report reasons for missing lesions, we distinguish perception and cognitive errors, and we comment on missing nodules, consolidations and infiltrative lung diseases. We provide tips to reduce our error rate, and in particular we comment on the importance of learning and applying key signs for optimizing the detection of abnormalities on both the frontal and the lateral views of the chest.
- Chest radiograph
- Chest radiology
- Thoracic imaging
To be aware of the actual risks of misdiagnosis when reading chest radiographs
To learn the best tips and tricks for reducing your error rate
To understand the limitations of chest radiographs compared with multi-detector CT
Missed lung lesions are one of the most frequent causes of malpractice issues [1,2,3]. Chest radiography plays an important role in the detection and management of patients with lung cancer, chronic airways disease, pneumonia and interstitial lung disease. Amongst all diagnostic tests, chest radiography is essential for confirming or excluding the diagnosis of most chest diseases. However, numerous lesions of a wide variety of disease processes affecting the thorax may be missed on a chest radiograph. For example, the frequency of missed lung carcinoma on chest radiographs can vary from 12 to 90%, depending on study design . Despite the lack of convincing evidence that screening for lung cancer with the chest radiograph improves mortality, chest radiography is still requested for this purpose. The chest radiograph will also help narrow a differential diagnosis, help to direct additional diagnostic measures and serve during follow-up. The diagnostic usefulness of the radiograph will be maximized by the integration of the radiological findings with the clinical features of the individual patient. In this chapter, we will review the more important radiological principles regarding missed lung lesions in a variety of common chest diseases, with a special focus on how correlation with multi-detector CT (MDCT) of missed lung lesions can help improve interpretation of the plain chest radiograph.
2.2 Reasons for Missed Lung Lesions
Conditions contributing to missed lung lesions, especially carcinomas, have been extensively studied [2, 4,5,6]. Poor viewing conditions, hasty visual tracking, interruptions, inadequate image quality and observer inexperience are amongst the most important [5, 7, 8]. Features of lesions themselves, when faced with nodules, such as location, size, border characteristics and conspicuity, also play a role . Missed lung nodules during initial reading of a chest radiograph are not uncommon. Missing a nodule which may represent malignancy will have adverse consequences on patient management, essentially through delayed diagnosis, which may carry medicolegal implications. A number of authors have explored the reasons why lesions are overlooked [9,10,11,12,13,14]. Specific studies have focussed on size , contrast gradient , conspicuity  and anatomic noise . Importantly, other types of errors, named systemic errors, can also occur  and include inappropriate orders or imaging utilization, procedure phase errors (patient identification, laterality, technique) and post-procedure phase errors (lighting conditions, transcription errors, communication failures).
One interesting study  examined the imaging features of non-small-cell lung carcinoma overlooked at digital chest radiography and compared general and thoracic radiologists’ performance for lung carcinoma detection. Frontal and lateral chest radiographs from 30 consecutive patients with lung carcinoma overlooked during initial reading and 30 normal controls were submitted to two blinded thoracic radiologists and three blinded general radiologists for retrospective review. The location, size, histopathology, borders, presence of superimposed structures and lesion opacity were recorded. Interobserver agreement was calculated, and detection performance between thoracic and general radiologists was compared. The average size of carcinomas missed by the thoracic radiologists was 18.1 mm (range 10–32 mm). The average size missed by general radiologists was 27.7 mm (range 12–60 mm). Seventy-one percent (5/7) of missed lesions were obscured by anatomical superimposition. Forty-three percent of lesions were located in the upper lobes, and 63% were adenocarcinomas. Compared with general radiologists, the lesions missed by thoracic radiologists tended to be smaller but also had significantly lower CT density measurements and more commonly had an ill-defined margin. The clinical stage of the overlooked lesions did not differ between the two groups (p = 0.480). The authors concluded that the lesion size, location, conspicuity and histopathology of lesions overlooked on digital chest radiography were similar to those missed on conventional film screen techniques.
The detection of carcinoma on a chest radiograph remains difficult with implications on patient management. Nowadays, it is still by far the most frequent cause of malpractice suits (42% of cases) . Whereas overlooking chronic airways disease, pneumonia and interstitial lung disease may not have the same potential medicolegal implications, the consequences for patient care could be critical.
We propose to review how correlation with multi-detector computerized tomography (MDCT) of missed lung lesions can help improve interpretation of the plain chest radiograph. During the course of clinical work, when reporting chest CT, whenever available, every effort should be made to review previous chest radiographs and their reports, thereby providing one of the best learning tools for chest radiograph interpretation.
Artificial intelligence will probably replace or modify our work as chest radiologists, minimizing detection errors and helping us to reduce our error rate. Convolutional neural networks have already been reported to provide a sensitivity of 97.3% and specificity of 100% in the detection of tuberculosis on chest X-rays .
A CT scan can be performed in patients with a negative chest radiograph when there is a high clinical suspicion of chest disease. CT scan, especially MDCT reconstructed with high-resolution algorithm and iterative reconstruction, is more sensitive than plain films for the evaluation of interstitial disease, bilateral disease, cavitation, empyema and hilar adenopathy. CT is not generally recommended for routine use because the data for its use in chronic airways disease and pneumonia are limited, the cost is high, and there is no evidence that outcome is improved. Thus, a chest radiograph is the preferred method for initial imaging, with CT scan reserved for further characterization (e.g., evaluation of pattern and distribution, detecting of cavitation, adenopathy, mass lesions or collections).
Many methods have been suggested for correct interpretation of the chest radiograph. There is no preferred scheme or recommended system. The clinical question should always be addressed. An inquisitive approach is always helpful and being aware of the areas where mistakes are made is essential. Hidden abnormalities can thus be looked for. The difficult “hidden areas” which must be checked are the lung apex, superimposed over the heart, around each hilum and below the diaphragm. We will concentrate on difficult areas such as lesions at the lung apices or bases or lesions adjacent to or obscured by the hila or heart. For a systematic approach, we will divide the review into three sections representing specific problems: missed nodules, missed consolidation and missed interstitial lung disease. Finally, we will illustrate some of the common signs that may help to detect lesions located in difficult anatomical areas of the chest.
Missing lesions is frequent.
Hidden areas are at highest risks for missing lesions.
Missing lesions is a frequent cause of medicolegal issues.
2.3 Specific Problems
Specific problems of missed lung lesions can be divided into missed nodules, missed consolidation and missed interstitial lung disease. In cases of a missed nodule or missed consolidation, the overlooked pathology may have been detected if special attention were paid to known “difficult areas”. The examples which follow will show how a side-by-side comparison of the chest radiograph and CT images improves our understanding of the overlooked lesion. There is no harm done by learning from one’s mistakes!
2.4 Missed Nodules
2.4.1 Nodular Lesions: Tumours
Nodular lesions are frequently due to lung cancer, which may be primary or secondary. Lung cancer is probably one of the most common lung diseases that radiologists encounter in practice. Berbaum formulated the concept that perception is better if you know where to look and what to look for . Our first example is that of a 53-year-old man who complained of pain in the right axilla for 4 months and underwent chest radiography. The postero-anterior and lateral radiographs were interpreted showing normal findings (Fig. 2.1a and b). The subsequent MDCT showed a right superior sulcus mass with rib destruction (Fig. 2.1c and d). Needle biopsy established a diagnosis of bronchogenic carcinoma (adenocarcinoma). Hindsight bias  with the information available from the MDCT makes the initial lesion extremely obvious. Careful scrutiny of both apices is essential when reporting a frontal chest radiograph.
Radiologic errors can be divided into two types : cognitive, in which an abnormality is seen but its nature is misinterpreted, and perceptual or the “miss”, in which a radiologic abnormality is not seen by the radiologist on initial interpretation. The perceptual type is estimated to account for approximately 80% of radiologic errors .
Our second patient illustrates the complexity of the detection of a lung nodule close to the hilum. A 77-year-old man with known prostate cancer underwent chest radiography for right upper quadrant abdominal pain (Fig. 2.2a and b). The radiographs were reported as normal. The coronal and sagittal reformats demonstrate the position of the nodule (Fig. 2.2c and d), which can be seen clearly with hindsight on the postero-anterior and lateral chest radiographs.
2.4.2 Nodular Lesions: Infections
Nodular lesions attributed to pulmonary infections are most often seen in nosocomial pneumonias and in immunocompromised patients. They may be caused by bacteria such as Nocardia asteroides and M. tuberculosis, septic emboli and fungi. Nocardia asteroides causes single or nodular infiltrates with or without cavitation. Invasive pulmonary aspergillosis (IPA), mucor and Cryptococcus neoformans may present with single or multiple nodular infiltrates, which often progress to wedge-shaped areas of consolidation. Cavitation (the “crescent sign”) is common later in the course of the infiltrate. In the appropriate clinical setting, CT may aid in the diagnosis of IPA by demonstrating the so-called halo sign. Figure 2.3 shows a 43-year-old woman with fever after a bone marrow transplant. The postero-anterior radiograph was interpreted as normal (Fig. 2.3a). With hindsight, a subtle infiltrate can be seen at the left apex. Conspicuity is lessened by the overlying clavicle and first rib. Axial CT image (Fig. 2.3b) shows nodular consolidation with crescentic cavitation (the “crescent sign”) and surrounding ground-glass infiltrate (the “halo sign”). These characteristic findings of IPA are best identified on CT.
Nodule location in hidden areas is the most frequent cause for missing nodules
Low nodule attenuation favours missing the lesion
Calcified nodules are easiest to detect but not clinically relevant
2.5 Missed Consolidation
2.5.1 Airspace Disease
Airspace disease is usually caused by bacterial infections. However, airspace disease can be seen in viral, protozoal, fungal infections and malignancy, typically brochioloalveolar carcinoma. Acute airspace pneumonia is characterized by a mostly homogeneous consolidation of lung parenchyma, well-defined borders, and does not typically respect segmental boundaries. An air bronchogram is very common. Progression to lobar consolidation may occur. As with lung nodules, whether consolidation is detected or missed on the plain chest radiograph may be determined by any combination the same factors of size, density, location and overlying structures. Location is a significant factor for missed consolidation. Consolidation in the middle lobe and both lower lobes can be difficult to diagnose, especially when only the postero-anterior view is obtained. Figure 2.4 shows a 46-year-old woman with cough and right-sided chest pain. The postero-anterior radiograph was interpreted as normal (Fig. 2.4a). Due to a clinical suspicion of pulmonary embolism, MDCT was requested, showing consolidation in the anterior segment of the right lower lobe. The coronal and sagittal reformats demonstrate the extent of the consolidation (Fig. 2.4b and c). There were no signs of pulmonary embolism on the contrast media study. A diagnosis of right lower lobe pneumonia was established, and the patient was treated successfully with antibiotics.
Chest radiography is the first recommended imaging test for the diagnosis of pneumonia. Chest radiography can diagnose pneumonia when an infiltrate is present and differentiate pneumonia from other conditions that may present with similar symptoms, such as acute bronchitis. The results of the chest radiograph may occasionally suggest a specific aetiology (e.g., a lung abscess) and identify a complication (empyema) or coexisting abnormalities (bronchiectasis, bronchial obstruction, interstitial lung disease). Chest radiography remains a valuable diagnostic tool in primary care patients with a clinical suspicion of pneumonia to substantially reduce the number of patients misdiagnosed. MDCT imaging is useful in patients with community-acquired pneumonia when there is an unresolving or complicated chest radiograph and at times in immunocompromised patients with suspected pulmonary infections. MDCT can help in differentiating infectious from non-infectious abnormalities. MDCT may detect empyema, cavitation and lymphadenopathy when the chest radiograph cannot. MDCT should be performed in immunocompromised patients with a clinical suspicion of pneumonia when the chest radiograph is normal. This is especially true when the early diagnosis of pneumonia is critical, as is the case with immunocompromised and severely ill patients.
Chest radiography is the first imaging test for the diagnosis of pneumonia.
Chest radiographs may help identify complications of pneumonia.
Hidden areas are the most frequent reasons for missing pneumonia.
2.6 Missed Interstitial Lung Disease
2.6.1 Diffuse (Interstitial or Mixed Alveolar-Interstitial) Lung Disease
Diffuse lung disease presenting with widely distributed patchy infiltrates or interstitial reticular or nodular abnormalities can be produced by a number of disease entities. An attempt is usually made to separate the group of idiopathic interstitial pneumonias from known causes, such as infections, associated systemic disease or drug related. The most common infectious organisms are viruses and protozoa. In general, the aetiology of an underlying pneumonia cannot be specifically diagnosed because the patterns overlap. It is beyond the aim of this chapter to discuss in detail the contribution of MDCT to the diagnosis of diffuse infiltrative lung disease. For over three decades, the development and then refinement of high-resolution computed tomography (HRCT) have resulted in markedly improved diagnostic accuracy in acute and chronic diffuse infiltrative lung disease. The chest radiograph remains the preliminary radiological investigation of patients with diffuse lung disease but is often non-specific. Pattern recognition in diffuse lung disease has been the subject of controversy for many years. Extensive disease may be required before an appreciable change in radiographic density, or an abnormal radiographic pattern can be detected on the plain chest radiograph. At least 10% of patients who are ultimately found to have biopsy-proven diffuse lung disease have an apparently normal chest radiograph. HRCT and now MDCT have become an integral component of the clinical investigation of patients with suspected or established interstitial lung disease. These techniques have had a major impact on clinical practice.
Chest radiography is less sensitive and less specific than MDCT.
If the chest radiograph is normal, MDCT may be indicated.
Chest radiographs may be helpful for the follow-up of ILD.
2.7 Key Signs for Reducing the Risk of Errors in CXRs
2.7.1 Deep Sulcus Sign
The deep sulcus sign (Fig. 2.5) is seen on chest radiographs obtained with the patient in the supine position . It represents lucency of the lateral costophrenic angle extending toward the abdomen. The abnormal deepened lateral costophrenic angle may have a sharp, angular appearance. When the patient is in the supine position, air in the pleural space (pneumothorax) collects anteriorly and basally within the nondependent portions of the pleural space; when the patient is upright, the air collects in the apicolateral location. If air collects laterally rather than medially, it deepens the lateral costophrenic angle and produces the deep sulcus sign. In Fig. 2.5, a deep sulcus sign is seen on the left, in addition to a continuous diaphragmatic sign, seen when air is seen between the diaphragm and the heart.
2.7.2 Spine Sign
On the normal lateral chest radiograph, the attenuation decreases (the lucency increases) as one progresses down the thoracic vertebral bodies. If the attenuation increases, locally or diffusely, there must be a posterior located lesion (Fig. 2.6). This lesion might not be seen on the frontal view, hidden by the heart or the hila. Interestingly, the positive predictive value of the spine sign is high (up to 97%) .
2.7.3 Silhouette Sign
In a chest X-ray, non-visualization of the border of an anatomical structure that is normally visualized shows that the area neighbouring this margin is filled with tissue or material of the same density (Fig. 2.6) . The silhouette sign is an important sign indicating the presence and the localization of a lesion.
2.8 Concluding Remarks
Despite the increasing use of CT imaging in the diagnosis of patients with chest disorders, chest radiography is still the primary imaging method in patients with suspected chest disease. The presence of an infiltrate on a chest radiograph is considered the “gold standard” for diagnosing pneumonia. Extensive knowledge of the radiographic appearance of pulmonary disorders is essential when diagnosing pulmonary disease. Chest radiography is also the imaging tool of choice in the assessment of complications and in the follow-up of patients with pulmonary diseases.
MDCT plays an increasing role in the diagnosis of chest diseases, especially in patients with unresolving symptoms. CT will aid in the differentiation of infection and non-infectious disorders. The role of CT in suspected or proven chest disease can be summarized as follows:
CT is valuable in the early diagnosis of chest disease, especially in patient groups in which an early diagnosis is important (immunocompromised patients, critically ill patients).
CT may help with the characterization of pulmonary disorders.
CT is an excellent tool in assessing complications of chest disease.
CT is required in the investigation patients with a persistent or recurrent pulmonary infiltrate.
A side-by-side comparison between the chest radiograph and MDCT when confronted with a missed lung lesion is very instructive. The radiologist should be able to understand the reasons for missing certain lesions. By adopting this inquisitive approach, both our cognitive and perceptual errors could be reduced.
Awareness of the dangers of systemic errors has become of upmost concern, as a result of the high examination volume and long shifts experienced by radiologists . Double readings and subsequent readings by subspecialists may become common practice, especially if medicine shifts toward physician payment based on quality or outcomes, rather than volume . Artificial intelligence will undoubtedly offer opportunities to improve our diagnostic accuracy, as systems will be developed as adjuncts to human cognition and perception .
Artificial intelligence generates fear about the future role of radiologists and their employment. Let us remember that although we analyse many images, we still decide on what imaging examinations should be prescribed and how they are performed best, we confer on difficult diagnoses, we discuss treatment plans with patients and we translate the conclusions of the research literature into real-life practice. If some of the more repetitive tasks can be handled safely by a computerized helper, radiologists will be able to focus on the rewarding ones, improving patient care and safety.
Be aware that missing lesions is frequent.
Always look at hidden areas.
Beware of satisfaction of search.
Take time to read the lateral view of the chest.
Learn the key radiologic signs to reduce your error rate.
Garland LH. Studies on the accuracy of diagnostic procedures. Am J Roentgenol Radium Therapy, Nucl Med. 1959;82(1):25–38.
Potchen EJ, Bisesi MA. When is it malpractice to miss lung cancer on chest radiographs? Radiology. 1990;175:29–32.
Baker SR, Patel RH, Yang L, Lelkes VM, Castro A. Malpractice suits in chest radiology: an evaluation of the histories of 8265 radiologists. J Thorac Imaging. 2013;28:388–91.
Quekel LG, Kessels AG, Goei R, et al. Miss rate of lung cancer on the chest radiograph in clinical practice. Chest. 1999;115:720–4.
Austin JH, Romney BM, Goldsmisth LS. Missed bronchogenic carcinoma: radiographic findings in 27 patients with a potentially resectable lesion evident in retrospect. Radiology. 1992;182:115–22.
Kim YW, Mansfield LT. Fool me twice: delayed diagnoses in radiology with emphases on perpetuated errors. AJR. 2014;202:465–70.. W.
Krupinski EA, Berger WG, Dallas WJ, et al. Searching for nodules: what features attract attention and influence detection? Acad radiol. 2003;10:861–8.
Busby LP, Courtier JL, Glastonbury CM. Bias in radiology: the how and why of misses and misinterpretations. Radiographics. 2018;38(1):236–47.
Kundel HL, Nodine CF, Krupinski EA. Searching for lung nodules. Visual dwell indicates locations of false-positive and false-negative decisions. Invest Radiol. 1989;24:472–8.
Samuel S, Kundel HL, Nodine CF, et al. Mechanism of satisfaction of search: eye position recordings in the reading of chest radiographs. Radiology. 1995;194:895–902.
Quekel LG, Goei R, Kessels AG, et al. Detection of lung cancer on the chest radiograph: impact of previous films, clinical information, double reading, and dual reading. J Clin Epidemiol. 2001;54:1146–50.
Tsubamoto M, Kuriyama K, Kido S, et al. Detection of lung cancer on chest radiographs: analysis on the basis of size and extent of ground-glass opacity at thin-section CT. Radiology. 2002;224:139–44.
Shah PK, Austin JH, White CS, et al. Missed non-small cell lung cancer: radiographic findings of potentially resectable lesions evident only in retrospect. Radiology. 2003;226:235–41.
Samei E, Flynn MJ, Peterson, et al. Subtle lung nodules: influence of local anatomic variations on detection. Radiology. 2003;228:76–84.
Kundel HL, Revesz G, Toto L. Contrast gradient and the detection of lung nodules. Investig Radiol. 1979;14:18–22.
Kundel HL. Peripheral vision, structured noise and film reader error. Radiology. 1975;114:269–73.
Kundel HL, Revesz G. Lesion conspicuity, structured noise, and film reader error. AJR. 1976;126:233–8.
Waite S, Scott JM, Legasto, et al. Systemic error in radiology. AJR. 2017;209:629–39.
Wu M-H, Gotway MB, Lee TJ, et al. Features of non-small cell lung carcinomas overlooked at digital chest radiography. Clin Radiol. 2008;63:518–28.
Lakhani P, Sundaram B. Deep learning at chest radiography: automated classification of pulmonary tuberculosis by using convolutional neural networks. Radiology. 2017;284:574–82.
Berbaum KS. Difficulty of judging retrospectively whether a diagnosis has been “missed”. Radiology. 1995;194:582–3.
Berlin L. Hindsight bias. AJR. 2000;175:597–601.
Berlin. Defending the “missed” radiographic diagnosis. AJR. 2001;176:317–22.
Berlin L, Hendrix RW. Perceptual errors and negligence. AJR. 1998;170:863–7.
Kong A. The deep sulcus sign. Radiology. 2003;228:415–6.
Medjek M, Hackx M, Ghaye B, De Maertelaer V, Gevenois PA. Value of the “spine sign” on lateral chest views. Br J Radiol. 2015;88(1050):20140378.
Algın O, Gökalp G, Topal U. Signs in chest imaging. Diagn Interv Radiol. 2011;17:18–29.
Hanna TN, Lamoureux C, Krupinski EA, et al. Effect of shift, schedule, and volume on interpretive accuracy: a retrospective analysis of 2.9 million radiologic examinations. Radiology. 2018;287:205–12.
Arenson RL. Factors affecting interpretative accuracy: how can we reduce errors ? Radiology. 2018;287:213–4.
Kahn CE. From images to actions: opportunities for artificial intelligence in radiology. Radiology. 2018;285:719–20.
Editors and Affiliations
© 2019 The Author(s)
About this chapter
Cite this chapter
Tack, D., Howarth, N. (2019). Missed Lung Lesions: Side-by-Side Comparison of Chest Radiography with MDCT. In: Hodler, J., Kubik-Huch, R., von Schulthess, G. (eds) Diseases of the Chest, Breast, Heart and Vessels 2019-2022. IDKD Springer Series. Springer, Cham. https://doi.org/10.1007/978-3-030-11149-6_2
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-11148-9
Online ISBN: 978-3-030-11149-6