Lung Cancer
Read Time: 6 mins

Advanced Lung Imaging – Magnetic Resonance Imaging and Artificial Intelligence

Authors: Katrina Mountfort
Senior Medical Writer, Touch Medical Media, UK
Copy Link
Published Online: Apr 11th 2019

While advances in radiotracers are increasing the utility of computed tomography (CT) and positron emission tomography (PET) in the imaging of lung disease, these techniques have limitations. Magnetic resonance imaging (MRI) is considered the most useful imaging modality in radiology, due to its ability to provide three-dimensional clear images, high spatial resolution and high contrast, and low acquisition time. In lung disease, the use of MRI offers the possibility of functional imaging without radiation burden, avoiding the need for CT scans in children, which can result in a small, but not negligible, increased rate of radiation-induced cancer.1 Clinical indications for MRI in lung disease include pulmonary vascular disease, especially pulmonary hypertension, airway diseases, especially cystic fibrosis; neoplastic disease, including staging of lung cancer as an alternative imaging modality; all paediatric indications (e.g., congenital anomalies); as well as follow-up examinations.2


While the use of MRI in the assessment of lung perfusion is well established,3 its use in the imaging of lung disease is has been limited because of its poor signal-to-noise ratio caused by the low proton density in the lung and the fast signal decay due to susceptibility artefacts at air-tissue interfaces.4 New technologies such as hyperpolarised gas MRI have overcome the limitations of proton-based MRI. In this technique, a noble gas such as 3He or 126Xe is hyperpolarised using specifically designed equipment. This hyperpolarised gas can be detected in MRI to produce images of ventilation or, using diffusion MRI protocols, to detect areas of structural damage in the lung.5 The technique has been shown promise in measuring disease progression and therapeutic response in patients with idiopathic pulmonary fibrosis,6 cystic fibrosis,7 chronic obstructive pulmonary disease (COPD)8 and asthma.9

Oxygen enhanced functional lung MRI, which exploits changes of the T1 times under normoxic and hyperoxic conditions related to a combination of ventilation, perfusion and diffusion capacity of the lung, is also emerging as a clinically useful technology. A recent prospective study (n=25) found that ultrashort echo time oxygen-enhanced MRI showed similar performance compared with hyperpolarised 3He MRI for quantitative measures of ventilation defects in patients with cystic fibrosis.10

Advances in technology are also expanding the utility of conventional MRI modalities. A new method for rebuilding static MRI images into moving videos has combined standard 2D MRI images of the chests of healthy volunteers to create ‘super-resolution’ videos, showing the lungs expanding and contracting.11 This technology could facilitate prediction of the location of a tumour, even as it moves, enabling more precise delivery of radiotherapy.

Recent studies have also suggested that MRI has potential as a screening method for lung cancer, with excellent sensitivity and specificity for nodules 6mm or greater.12,13 A mathematical model found that that MRI’s superior specificity for characterizing solid nodules, compared with CT, could result in a significant decrease in downstream workup costs. The authors concluded that lung MRI has the potential to be a cost-effective alternative to low-dose CT for lung cancer screening.14

The introduction of national lung cancer screening programmes is resulting in the generation of unprecedented amount of chest CT scans, which need interpretation by radiologists. Computer-aided diagnosis systems can greatly improve the efficiency and cost-effectiveness of this process. In recent years, the use of artificial intelligence (AI) has vastly improved the interpretation of complex data.15 Machine learning, i.e. teaching a computer how to arrive at a particular conclusion by feeding it large quantities of data, has been used to determine whether a single lung nodule was malignant,16 By contrast, deep learning involves teaching the computer to recognise the features of an image by layering a series of algorithms on top of each other to create an artificial neural network, which mimics to construction of the brain. A deep learning system has been developed that automatically classifies all nodule types relevant for nodule workup. The system processes raw CT data containing a nodule without the need for any additional information such as nodule segmentation or nodule size The system has been shown to be superior to classical machine learning approaches in terms of classifying nodules and is within the inter-observer variability among four experienced human observers.17

The use of AI algorithms may also reduce false positive rates in lung cancer screening, according to the findings of a new study. At present, among the 24% of positive screens requiring follow-up, 96% are false positives, resulting in costly and potentially harmful follow-up scans and biopsies.18 A machine learning algorithm, the Lung Cancer Causal Model (LCCM), has been developed and validated in a cohort of 126 subjects. This was able to identify 30% of benign nodules without risk of misclassifying cancer nodules, a substantial improvement over existing methods.19

Lung cancer is not the only condition in which AI can improve diagnostic accuracy. An algorithm called CheXNeXt has been trained using 112,000 X-rays. A panel of three radiologists then reviewed a different set of 420 X-rays, one by one, for 14 different pathologies. For 10 diseases, the algorithm performed as well as radiologists; for 3, it underperformed compared with radiologists; and in the case of pneumonia, the algorithm outdid the experts, diagnosing the disease with 22% greater accuracy than radiologists working alone and reducing errors by 33%. This algorithm is being further developed and will progress to in-clinic testing.20

Pulmonary function tests (PFT) provide large amounts of numerical data and their patterns can be hard for the human eye to and recognise. A recent study trained an algorithm using historical data from 1430 patients from 33 Belgian hospitals. Next, 120 pulmonologists from 16 European hospitals made 6000 interpretations of PFT data from 50 randomly selected patients. The AI-based software also evaluated the same data. Interpretation of the PFTs by the pulmonologists matched guidelines from the European Respiratory Society and the American Thoracic Society in 74% of cases (56–88%), but the AI algorithm interpretations matched the guidelines exactly (100%). Pulmonologists were able to correctly diagnose the primary disease in 45% of cases (24–62%), while the AI gave a correct diagnosis in 82% of cases.21

In summary, advances in technology have has brought MRI to the threshold of broad clinical application in lung disease. In addition, the use of AI promises to accelerate the productivity of radiologists and potentially improve the accuracy of screening and diagnostic tests. These technologies are likely to become invaluable in helping doctors and patients in all aspects of therapeutic decision-making.


1. Niemann T, Colas L, Roser HW, et al. Estimated risk of radiation-induced cancer from paediatric chest CT: two-year cohort study. Pediatr Radiol. 2015;45:329–36.

2. Kauczor H-U, Ley-Zaporozhan J, Ley S. Imaging of Pulmonary Pathologies: Focus on Magnetic Resonance Imaging. Proc Am Thor Soc. 2009;6:458–63.

3. Ley S, Ley-Zaporozhan J. Pulmonary perfusion imaging using MRI: clinical application. Insights Imaging. 2012;3:61–71.

4. Ciet P, Harm AWM, Tiddens PA, et al. Magnetic resonance imaging in children: common problems and possible solutions for lung and airways imaging. Pediatric Radiology. 2015;45:1901–15.

5. Kern AL, Vogel-Claussen J. Hyperpolarized gas MRI in pulmonology. Br J Radiol. 2018;91:20170647.

6. Wang JM, Robertson SH, Wang Z, et al. Using hyperpolarized (129)Xe MRI to quantify regional gas transfer in idiopathic pulmonary fibrosis. Thorax. 2018;73:21–8.

7. Kirby M, Svenningsen S, Ahmed H, et al. Quantitative evaluation of hyperpolarized helium-3 magnetic resonance imaging of lung function variability in cystic fibrosis. Acad Radiol. 2011;18:1006–13.

8. Doganay O, Matin T, Chen M, et al. Time-series hyperpolarized xenon-129 MRI of lobar lung ventilation of COPD in comparison to V/Q-SPECT/CT and CT. Eur Radiol. 2018; doi: 10.1007/s00330-018-5888-y. [Epub ahead of print].

9. Svenningsen S, Eddy RL, Lim HF, et al. Sputum Eosinophilia and Magnetic Resonance Imaging Ventilation Heterogeneity in Severe Asthma. Am J Respir Crit Care Med. 2018;197:876–84.

10. Zha W, Nagle SK, Cadman RV, et al. Three-dimensional Isotropic Functional Imaging of Cystic Fibrosis Using Oxygen-enhanced MRI: Comparison with Hyperpolarized (3)He MRI. Radiology. 2019;290:229–37.

11. Freedman JN, Collins DJ, Gurney-Champion OJ, et al. Super-resolution T2-weighted 4D MRI for image guided radiotherapy. Radiother Oncol. 2018;129:486–93.

12. Meier-Schroers M, Homsi R, Gieseke J, et al. Lung cancer screening with MRI: Evaluation of MRI for lung cancer screening by comparison of LDCT- and MRI-derived Lung-RADS categories in the first two screening rounds. Eur Radiol. 2019;29:898–905.

13. Meier-Schroers M, Homsi R, Skowasch D, et al. Lung cancer screening with MRI: results of the first screening round. J Cancer Res Clin Oncol. 2018;144:117–25.

14. Kim A. MRI shows promise for lung cancer screening. Available at: (accessed 8 April 2019).

15. Hosny A, Parmar C, Quackenbush J, et al. Artificial intelligence in radiology. Nat Rev Cancer. 2018;18:500–10.

16. McWilliams A, Tammemagi MC, Mayo JR, et al. Probability of cancer in pulmonary nodules detected on first screening CT. N Engl J Med. 2013;369:910–9.

17. Ciompi F, Chung K, van Riel SJ, et al. Towards automatic pulmonary nodule management in lung cancer screening with deep learning. Sci Rep. 2017;7:46479.

18. Aberle DR, Adams AM, Berg CD, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011;365:395–409.

19. Raghu VK, Zhao W, Pu J, et al. Feasibility of lung cancer prediction from low-dose CT scan and smoking factors using causal models. Thorax. 2019;doi: 10.1136/thoraxjnl-2018-212638. [Epub ahead of print].

20. Armitage H. Artificial intelligence rivals radiologists in screening X-rays for certain diseases. Available at: (accessed 8 April 2019).

21. Topalovic M, et al. Artificial intelligence improves experts in reading pulmonary function tests. Presented at: the European Respiratory Society International Congress. Paris, France, 15–19 September 2018. Abstract PA5290.


  • Copied to clipboard!
    accredited arrow-down-editablearrow-downarrow_leftarrow-right-bluearrow-right-dark-bluearrow-right-greenarrow-right-greyarrow-right-orangearrow-right-whitearrow-right-bluearrow-up-orangeavatarcalendarchevron-down consultant-pathologist-nurseconsultant-pathologistcrosscrossdownloademailexclaimationfeedbackfiltergraph-arrowinterviewslinkmdt_iconmenumore_dots nurse-consultantpadlock patient-advocate-pathologistpatient-consultantpatientperson pharmacist-nurseplay_buttonplay-colour-tmcplay-colourAsset 1podcastprinter scenerysearch share single-doctor social_facebooksocial_googleplussocial_instagramsocial_linkedin_altsocial_linkedin_altsocial_pinterestlogo-twitter-glyph-32social_youtubeshape-star (1)tick-bluetick-orangetick-red tick-whiteticktimetranscriptup-arrowwebinar Sponsored Department Location NEW TMM Corporate Services Icons-07NEW TMM Corporate Services Icons-08NEW TMM Corporate Services Icons-09NEW TMM Corporate Services Icons-10NEW TMM Corporate Services Icons-11NEW TMM Corporate Services Icons-12Salary £ TMM-Corp-Site-Icons-01TMM-Corp-Site-Icons-02TMM-Corp-Site-Icons-03TMM-Corp-Site-Icons-04TMM-Corp-Site-Icons-05TMM-Corp-Site-Icons-06TMM-Corp-Site-Icons-07TMM-Corp-Site-Icons-08TMM-Corp-Site-Icons-09TMM-Corp-Site-Icons-10TMM-Corp-Site-Icons-11TMM-Corp-Site-Icons-12TMM-Corp-Site-Icons-13TMM-Corp-Site-Icons-14TMM-Corp-Site-Icons-15TMM-Corp-Site-Icons-16TMM-Corp-Site-Icons-17TMM-Corp-Site-Icons-18TMM-Corp-Site-Icons-19TMM-Corp-Site-Icons-20TMM-Corp-Site-Icons-21TMM-Corp-Site-Icons-22TMM-Corp-Site-Icons-23TMM-Corp-Site-Icons-24TMM-Corp-Site-Icons-25TMM-Corp-Site-Icons-26TMM-Corp-Site-Icons-27TMM-Corp-Site-Icons-28TMM-Corp-Site-Icons-29TMM-Corp-Site-Icons-30TMM-Corp-Site-Icons-31TMM-Corp-Site-Icons-32TMM-Corp-Site-Icons-33TMM-Corp-Site-Icons-34TMM-Corp-Site-Icons-35TMM-Corp-Site-Icons-36TMM-Corp-Site-Icons-37TMM-Corp-Site-Icons-38TMM-Corp-Site-Icons-39TMM-Corp-Site-Icons-40TMM-Corp-Site-Icons-41TMM-Corp-Site-Icons-42TMM-Corp-Site-Icons-43TMM-Corp-Site-Icons-44TMM-Corp-Site-Icons-45TMM-Corp-Site-Icons-46TMM-Corp-Site-Icons-47TMM-Corp-Site-Icons-48TMM-Corp-Site-Icons-49TMM-Corp-Site-Icons-50TMM-Corp-Site-Icons-51TMM-Corp-Site-Icons-52TMM-Corp-Site-Icons-53TMM-Corp-Site-Icons-54TMM-Corp-Site-Icons-55TMM-Corp-Site-Icons-56TMM-Corp-Site-Icons-57TMM-Corp-Site-Icons-58TMM-Corp-Site-Icons-59TMM-Corp-Site-Icons-60TMM-Corp-Site-Icons-61TMM-Corp-Site-Icons-62TMM-Corp-Site-Icons-63TMM-Corp-Site-Icons-64TMM-Corp-Site-Icons-65TMM-Corp-Site-Icons-66TMM-Corp-Site-Icons-67TMM-Corp-Site-Icons-68TMM-Corp-Site-Icons-69TMM-Corp-Site-Icons-70TMM-Corp-Site-Icons-71TMM-Corp-Site-Icons-72