Liver - LI-RADS

Liver Imaging Reporting And Data System

Frederieke Elsinger, Christopher Lunt, Alison Harris and Silvia Chang

Vancouver General Hospital

This article is still work in progress.
It will be announced on the home page when it is ready.

The Liver Imaging Reporting and Data System (LI-RADS) is a classification system which is used in patients with liver cirrhosis and chronic hepatitis B virus infection (HBV) who  have an increased risk of hepatocellular carcinoma (HCC).

The LI-RADS categories reflect the probability of HCC and are based on the typical CT or MR-findings in HCC.


LI-RADS - Major features

There are five major features which are typically seen in HCC in patients with livercirrhosis and  chronic hepatitis B virus infection (fig).

  1. Arterial phase hyperenhancement (APHE)
    APHE is non-rim arterial hyperenhancement of lesion which is greater than the enhancement of the surrounding liver. Rim enhancement is not a feature of HCC.
  2. Non-peripheral washout
    Decrease in attenuation or intensity from earlier to later phase, resulting in hypoenhancement in the portal venous or delayed phase.
  3. Capsule
    Smooth, uniform border surrounding all or most of an observation.
  4. Size 
    A large lesion has a greater chance of being a HCC than a small lesion.
  5. Threshold growth
    Threshold growth is increase in size of 50% or more within 6 months time during follow-up imaging. 
Click to enlarge

LI-RADS categories

The term observation is used when it is not sure, whether something is a lesion or a pseudolesion.

The LI-RADS categories reflect the probability of an observation being a HCC and ranges from LR-1 (definitely benign) to LR-5 (definitely HCC).

Use LI-RADS as follows:

  1. First determine whether there is enhancement.
  2. Then look at the type of enhancement.
    Non-rim hyperenhancement is more suspicious than hypo- or isoenhancement.
  3. Then look at the size of the lesion.
  4. Finally look for additional typical features of HCC like enhancing capsule, non-peripheral washout and growth of the lesion.

Additional categories are:

  • LR-TIV when there is tumor in a vein.
  • LR-M when a lesion is probably malignant but not HCC.
  • LR-TR when we are dealing with a treated HCC.

The most interesting part of the table is when we have to decide if an observation has to be placed in category LR-3, which is intermediate probability or in LR-4, which is probable HCC or in category LR-5, which is definitely HCC.

As mentioned above, first determine if there is enhancement and if it is non-rim APHE.
Then look at the size of the observation and finally count the number of additional major features.

Mass versus Pseudolesion

LI-RADS uses the term ‘observations’ to describe focal abnormalities which are distinct from the background liver parenchyma.
It is preferred not to use the word ‘lesion’ or ‘nodule’ as some of these abnormalities do not represent true lesions but are the result of a perfusion alteration, a hypertrophic pseudomass or an artifact. 

When to use LI-RADS

LI-RADS is used in patients with cirrhosis and chronic HBV infection.

LI-RADS should not be used in patients with cirrhosis caused by vascular disorders (see table).
In these patients the formation of benign hyperplastic nodules may resemble HCC on imaging and cause false positive diagnoses.

Major features

Typical HCC with APHE and washout.

Arterial Phase non-rim Hyperenhancement (APHE)

Here an image in the late arterial phase in a patient with cirrhosis.
There is an observation with non-rim hyperenhancement.
In a late phase there is washout.
These are typical features of HCC.

Typical HCC with APHE, washout and enhancing capsule

Washout - Capsule

This is another patient with an enhancing lesion and washout.
Note also the enhancing capsule on the delayed phase.

A capsule is one of the major features of HCC and can be complete or partial.
A capsule should always be included within the measurement of the lesion.

LEFT: Hyperenhancing observation detected in segment 5. RIGHT: follow up 3 months later shows growth.

Size - Threshold growth

Size also determines in which category a lesion is placed.
The bigger the lesion, the higher the chance that it is a HCC.

Threshold growth is also an important finding.
It is defined as more than 50% growth in less than 6 months

The images show an observation in segment 5 of the liver demonstrating arterial hyperenhancement.
The lesion has grown from 8 mm to 21 mm in 3 months, which means that there is threshold growth.

Ancillary features

Features favoring HCC

Ancillary features are findings that are helpful for detection improvement, increase in confidence or category adjustment. 
These features are not obligatory and can be used at the radiologist’s discretion. 
In case of category adjustment, observations can only be upgraded or downgraded one category.
However you are not allowed to upgrade from LR-4 to LR-5, because these ancillary features lack sufficient specificity for HCC to allow for an LR-5 upgrade.

Non-enhancing capsule is a feature of a capsule surrounding an observation not appearing as an enhancing rim.
Nodule in nodule is a nodule within a larger nodule with different imaging characteristics
Mosaic structure is randomly distributed compartments or nodules within an observation, usually with different imaging features.
Blood products in mass is intralesional or perilesional hemorrhage in the absence of biopsy or trauma.
Fat in mass is excess of fat within the whole or part of the mass, more than in adjacent liver. Can be large extracellular fat or intracellular.

Nodule in nodule

The images show a fat containing lesion with arterial hyperenhancement.
Within this lesion there is a nodule (arrow) with wash out and a capsule.

The lesion was classified as LR-5.

Mosaic architecture

The image shows an observation with hyperenhancement in the late arterial phase in a patient with cirrhosis.
Notice the mosaic architecture.

IP and OOP image

Blood products in mass

Study the CT-image.
What are the findings?

The findings are high signal in a mass in segment II both on an in-phase image aswell as on the out-of-phase image (arrow).

This represents internal hemorrhage and in the absence of biopsy or trauma, is a feature that favors the diagnosis of HCC.

HCC classified as LI-RADS 5.

Fat in mass

Excess of fat in the whole or part of a mass is an ancillary findings that favors the diagnosis of HCC.

Same image as shown earlier.
What are the findings?

The findings are:

  • Chirhosis: irregular contour of the liver, large spleen and ascites.
  • Observation with non-rim and macroscopic fat in segment VII of the liver (-50 HU, yellow arrow).

This was classified as LI-RADS 5.

Study the CT-image.
What are the findings?

The findings are:

  • non-rim APHE in segment VI of the liver
  • Washout on PV phase 
  • Enhancing capsule
  • Microscopic fat on IP/OOP sequences

Continue with next image...

Based on the findings on the images shown above, the lesion is classified as LI-RADS 5 because it is larger that 20mm and shows hyperenhancement and washout with a capsule.

The ancillary features favoring HCC are intracellular fat on IP/OOP imaging and diffusion restriction.
Now we can not go any higher than LI-RADS 5, but the extra features will give us extra confidence in the diagnosis.

Diffusion restriction

In this patient with cirrhosis the MR-images show an arterially enhancing observation (< 2 cm) in the right lobe (arrow).
As there is no washout or any other major feature observed this should be classified as a LR-3 lesion.
However due to the ancillary finding of diffusion restriction this observation can be upgraded to a LR-4.

This is a difficult topic, since diffusion restriction can also be seen in non-HCC tumors.
When in doubt between LR-4 and LR-M sometimes a biopsy is performed.

Favoring non-HCC Malignancy

The table shows an overview of ancillary features favoring not HCC specific malignancy.

Features favoring benignity

The table shows an overview of benign ancillary features.

Siderotic nodules

Cirrhotic liver containing multiple small siderotic nodules which are of low signal on the T2-weighted images due to the T2 shortening of iron. There is also signal drop on the T1 inphase sequence compared to the opposed sequence and absence of contrast enhancement on the subtraction sequences.

LI-RADS categories

Non enhancing lesions in a cirrhotic liver, compatible with cysts, LR-1

LI-RADS 1 - definitely benign

Observations in this category are definite benign. 

Examples of LR-1 lesions are cysts or lesions that show spontaneous disappearance. 

The images show lesions that show no enhancement in both late arterial, portal venous and delayed phase.

Typical haemangioma in a patient with livercirrhosis.

LI-RADS 2 - probably benign

LR-2 observations are probably benign.
Of all LR-2 lesions about 16% are HCC and 18% are malignant. 

Examples are perfusions alterations (shunts), hemangiomas, hepatic fat deposition or fat-sparing, hypertrophic pseudo mass, confluent fibrosis or scar. 

Distinct nodules < 20 mm without malignant features (see section of ancillary malignant features) can be categorized as LR-2.

Examples are nodules which are T1 hyperintense, T2 hypointense, siderotic or hepatobiliary phase hyperintense.

Continue with the MR of this patient.

MRI-images of the same patient.

The enhancement of this lesion follows the enhancement of the blood pool, which is typical for a hemangioma.

LI-RADS 3 - intermediate probability

LR-3 observations vary from benign lesions to dysplastic nodules to HCC. 
Many LR-3s are vascular pseudolesions. 
Of all LR-3 lesions approximately 37% are HCC and 39% are malignant. 

Lesions that are placed in category LI-RADS 3 are:

  • Nodules with features of focal nodular hyperplasia or hepatic adenoma. 
  • Nodules of < 20mm without major features but with one or more ancillary findings of malignancy such as intralesional fat, T2 hyperintensity, diffusion restriction and HB phase hypointensity. 
  • Nodules above 20 mm and without major or ancillary features.

Study the MR-images. What are the findings?
How would you score LI-RADS.

The findings are:

  • Post contrast imaging shows a lesion with non-peripheral hyperenhancement.
  • lesion < 2cm
  • No washout on the portal venous or delayed phase.

The lesion was classified as LI-RADS 3.

Here another small hyperenhancing lesion without any additional feature like washout, capsule or threshold growth.

This was also classified as LI-RADS 3.

LI-RADS 4 - probably HCC

Of all LR-4 observations about 74% are HCC and 81% are malignant. 
The categorizing of an observation as LR-4 depends on the size of the lesion, the presence of APHE and the amount of additional major features (see LIRADS table).
If an arterial enhancing lesion measures between 10-19 mm and has one additional major feature, the assigned category depends on which type of feature is present. 
If capsular enhancement is noticed the lesion should be described as LR-4. 
If there is wash out or threshold growth the category changes to LR-5.

LI-RADS 5 - definitely HCC

Study the MR-images.
What are the findings?
How would you score LI-RADS.

The findings are:

  1. lesion > 20mm
  2. non-rim APHE
  3. washout

This is a LR-5 lesion.
There is one ancillary finding and that is the diffusion restriction (arrow).
This is a helpful finding and we will discuss that later.

Of all LR-5 lesions 95% are HCC and 98% are malignant.
In patients with concurrent extra-hepatic malignancy the positive predictive value of LR-5 for HCC drops, especially if the primary tumor is hypervascular. 
When in doubt, categorizing a lesion as LR-M might be more appropriate in this group of patients.

The images show an arterially enhancing lesion with washout in segment VI. suspicious of HCC: LR-5. 

Note that the arterial enhancement is faint because the scanning is too early for late arterial phase scanning.
Additional small lesions can therefore be easily missed.

LR-tumor in vein

LR-TIV - tumor in vein

The classification LR-TIV should be applied when there is unequivocal soft tissue within a vein, regardless if an associated mass is seen or not.
Almost always the venous invasion by tumor is related to a HCC.
In a small percentage however this is caused by other, non-HCC malignancies. 
Additional clues of possible venous invasion by tumor can be the presence of an occluded vein with ill-defined margins or diffusion restriction. 

An occluded vein contiguous with a malignant parenchymal mass should also raise suspicion of tumor invasion and should be scrutinized for enhancing soft tissue.

LR-tumor in vein

The images show a mild rim-like arterially enhancing lesion in segment V with washout.
A linear area of hypoenhancement is seen extending from the mass which is suspicious of tumor in vein (yellow arrow). 

Kunnen we hier nog iets zeggen over de lesie zelf en deze scoren?

A follow up CT was done, showing severe progression of tumor as well as vascular involvement of the anterior right portal vein (white arrow). 

LR-M - malignant

The category of LR-M should be applied to malignant appearing lesions that do not have the typical characteristics of HCC. 

Non-HCC malignant lesions either show:

  • Targetoid appearance: peripheral APHE and washout and delayed central enhancement.
  • Non-targetoid with an infiltrative growth pattern. 
  • Marked diffusion restriction or necrosis. 

Categorizing a lesion as LR-M does not mean the diagnosis of  HCC is excluded.
Of all LR-M lesions 2/3 are non-HCC malignancies like intrahepatic cholangiocarcinomas (CCA) or combined HCC-CCA and about 5% are benign.
This small percentage of benign malignant appearing lesions usually represent sclerosing hemangiomas or abscesses.

Heb je een voorbeeld van zo'n benigne lesie die lijkt op HCC?

Click to enlarge

Tumor markers

Evaluation of circulating tumor biomarkers such as AFP and CA19-9 can be helpful to refine the differential diagnosis.

The images show a large rim enhancing mass in the late arterial phase in a patient with chronic hepatitis.
In the portal venous phase there is progressive peripheral enhancement. 

Pathology diagnosis confirmed this was not a HCC but a cholangiocarcinoma

The images show a large, heterogeneous enhancing lesion in segment II.
There is peripheral enhancement in the portal venous phase. There is high signal on T1 IP and OOP imaging and diffusion restriction. 

This lesion was resected and pathology showed a mixed HCC-iCCA.


LI-RADS treatment response - LRTR

Evaluation of treatment response

A different algorithm was created for the categorization of treated lesions. 
Examples of locoregional therapies are radiofrequency ablation, percutaneous ethanol ablation, cryoablation, microwave ablation, transarterial embolization or chemoembolization, doxirubicin-eluted bead chemoembolization, transarterial radioembolization and external beam radiotherapy.

These lesions are either characterized as non evaluable, nonviable, equivocal or viable. 
Treatment related parenchymal perfusional changes may mimic or obscure residual tumor, potentially leading to false positive or false negative assessment of viability.
If there is any uncertainty between two categories, the one reflecting lower certainty should be applied.

LI-RADS does not apply for systemic treatment.

LRTR non viable

If an observation shows no residual major or ancillary features or has completely disappeared it can be characterized as non-viable. 

If typical post-treatment features are noted in the area of treatment and there are no signs of residual malignancy, a lesion can be considered as non viable.

the images show an ablation defect in segment V adjacent to gallbladder  and hepatic vein without residual areas of arterial enhancement or washout.
This was classified as  LRTR non viable.

The image in the late arterial phase show normal post treatment pattern with rim-like hyperenhancement of the surrounding liver parenchyma due to hyperemia after DEB-TACE (arrow). 

LRTR non viable

LRTR equivocal

The enhancement pattern is atypical for treatment-specific expected enhancement and otherwise not meeting criteria for probably or definitely viable.
For some treatments early post-treatment enhancement patterns may not reliably differentiate viable from non viable tumor and therefore the most appropriate category in this early post-treatment period may be LRTR equivocal.

Area of arterial enhancement without washout in a segment 5 ablation defect (red arrow), LRTR-equivocal. 

LRTR viable

If a new tumor arises adjacent to a treated lesion this observation should be assigned a new non-treated LIRADS category.

If a new observation is noted at the surgical margin it should be assigned a category according to the treated lesion algorithm.

Large lesion treated with TACE with residual areas of mild arterial enhancement and washout due to untreated supply via a phrenic artery (black arrow); LRTR-viable. Note there is irregularity of the peritoneal fat anteriorly (red arrow), in keeping with capsular rupture. Another example (sices at two different levels) below of a patient with residual eccentric enhancement and washout after treatment; LRTR-viable. 

When measuring viable tumor only the largest continuous area of enhancement or washout should be measured (not traversing non enhancing area).
If residual enhancement  within a lesion is nodular, the largest nodule should be reported.

Previous segment II resection with viable tumor at the resection margin.
Note the area of arterial enhancement and washout adjacent to the surgical clips (arrows).
This is  LRTR-viable.



Early aterial phase (left). Late arterial phase (right).

CT protocol

Required images for CT are late-arterial, portal venous and delayed phase.
Precontrast images are recommended after locoregional treatment.

Late arterial phase
This phase refers to the hepatic arterial phase in which the hepatic artery and branches are fully enhanced and the hepatic veins are not yet enhancing. 

HCC usually enhances more strongly in the late arterial phase showing early enhancement of the portal vein and is therefore preferred over the early arterial phase, when there is no enhancement of PV and poor enhancement of the HCC.

Notice poor enhancement of liver and portal vein (white arrow) in the early arterial phase and good enhancement of both hepatic artery (white arrow) and portal vein (green arrow) in the late arterial phase.

Portal venous phase

Portal venous phase
There is complete enhancement of the portal veins with antegrade filling of hepatic veins.
In this phase the normal liver parenchyma is usually at its peak of enhancement.
Hypovascular lesions like most metastases are best detected in this phase, but hypervascular lesions are poorly seen because there is poor contrast between the good enhancement of the liver parenchyma and the hypervascular lesion.

Delayed phase

Delayed phase

Portal and hepatic veins are enhanced but less than in the portal venous phase. 
Liver parenchyma is enhanced but also less than in PV phase.
This phase is typically acquired 2-5 min after injection.

This phase is helpful in detecting wash out of HCC and in detecting sclerotic lesions like cholangiocarcinoma or sclerotic metastases like sometimes seen in breast cancer.

Hepatobiliary phase study: diffuse uptake of contrast by normal liver parenchyma with hypointense appearance of the vessels and contrast excreted within the bile ducts.

MRI protocol

The required sequences for MRI are: 

  • Unenhanced T1 IP and OP imaging
  • T2 (with or without fat suppression) 
  • Multiphase T1 pre and post contrast. 
  • Optional sequences: diffusion-weighted and subtraction imaging.

Transitional phase
This phase is acquired with a hepatobiliary contrast agent after the extracellular phase but before the hepatobiliary phase.
Liver vessels and hepatic parenchyma are of similar intensity. Typically acquired 2-5 min after injection.

Hepatobiliary phase
This phase is acquired with a hepatobiliary agent.
Liver parenchyma is hyperintense to the hepatic vessels. There is excretion of contrast in the biliary system.
Typically acquired about 20 min after injection of gadoxetate or 1-3 hours after gadobenate.

Hypervascular metastases

Multiple heterogenous lesions with mild peripheral enhancement in the arterial phase.
Some lesions are mildly hypodense compared to adjacent liver parenchyma on the portal phase and larger lesions have a necrotic center.
Multiplicity and targetoid enhancement pattern are not typical for HCC and suggestive of metastatic disease.
In this case of a neuroendocrine tumor. 

MRI of a steatotic liver containing multiple lesions which are hyperenhancing in the arterial and portal venous phase.
Patient was known with a neuroendocrine tumor of the pancreas.
Findings are in keeping with hypervascular metastases.