Acid-fastness in microbiology: The versatility of Ziehl–Neelsen staining technique, its modifications, and integration of artificial intelligence – A brief review

INTRODUCTION

In the year 1882, Mycobacterium tuberculosis, an etiological agent of tuberculosis (TB), was described and isolated first time by Robert Koch, and after that, Paul Ehrlich demonstrated the acid–fast (AF) nature of bacilli using aniline dyes. Franz Ziehl in 1890s used phenol a carbolic acid and Friedrich Neelsen used basic fuchsin (carbol fuchsin) as a primary stain, and thus, they modified the method used by Paul Ehrlich and the technique is now known as the Ziehl–Neelsen (ZN) method.^[1] ZN staining method is also known as AF staining, known to identify AF structures, especially Mycobacterium as well as Actinomyces spp. Acid fastness is a physical property that gives a bacterium the ability to resist decolorization by acids during staining procedures.^[2] As heating of the stain is required in this method, it is also known as the hot method.

AF bacteria have huge amounts of mycolic acid present in their cell wall, because of which they are resistant to stains such as Gram stain. In ZN stain, carbol fuchsin gives each cell a red color, which is then removed by strong acid such as acid alcohol or sulfuric acid (20%). Resistance to decolorization by acid–alcohol (i.e., “acidfastness”) is associated with the mycolic acid–arabinogalactan moieties that constitute the bulk of cell wall materials external to the peptidoglycan layer. The soluble lipids also contribute but do not determine the AF properties of AF cells this indicating that the total lipid content (mainly mycolic acid) of the cell wall is responsible for the AF staining property. The non-AF structures/cells do not have a thick lipid layer or mycolic acid that can retain carbol fuchsin when decolorized . Therefore, when exposed to methylene blue (counterstain), they take it up and appear blue under the microscope, while AF structures retain carbol fuchsin and appear red. AF structures retain carbol fuchsin and look red. ^[3,4]

MATERIALS AND METHODS

A search was conducted on PubMed and Google Scholar to identify the published scientific literature related to the topic. Titles, abstracts, and keywords were searched using a combination of terms, such as ZN staining, AF staining, and artificial intelligence (AI). The databases were searched from inception to 10 May 2025. The research papers, review articles, and papers related to topic were identified and chosen for this review.

STAINING PROCEDURE

STEPS OF STAINING

1. Smear is flooded with carbol fuchsin (primary stain) and gently heated until steam rises (do not boil as it may alter the results), for 5–10 min. Allow stain to penetrate. This heating enhances the penetration of the dye inside the bacterial cell (i.e., the ZN method). In this step, make the entire smear take red color. Incorporation of detergent into the dye (i.e., the Kinyoun method) also enhances the penetration of the dye inside AF cells but this method is obsolete now days^[3,5,6]

2. The second step after stain fixation is the washing of smear under gently tap water to remove excess stain

3. The step third in this procedure is decolorization of the smear with strong acid such as 25% H₂SO₄. The decolorization is applied on the smear for 2–3 min, or until smear turns light pink. The decolorizer decolorizes all non-AF structures completely, thus leaving only AF structures of red color behind^[7]

4. The smear is again washed in a stream of water

5. Methylene blue (counterstain) is then applied on the smear for 1 min; this creates visual contrast during microscopy^[3,5]

6. Again, the smear is washed and air dried after which observation is made at 100× oil immersion lens.

Result interpretation

The slide is observed for bright red/pink bacilli or AF structures on a blue background at 100 × oil immersion lens as shown in Figure 1 and the result interpretation is done as per the National Tuberculosis Elimination Program guidelines as shown in Table 2.^[7]

Close

Figure 1:

Red colour arrows indicating rod-shaped bacilli (red arrows) against a blue background in sputum smear stained using the ZN method. Scale bar = 10 µm; magnification = 4000×. ZN: Ziehl–Neelsen staining (Source: Wikimedia Commons).

Export to PPT

Table 2: Ziehl–Neelsen stain grading system as per the National Tuberculosis Elimination Program, India, used for reporting AFB in sputum smear microscopy

Number of AFB/100 OIF	Grading	Result
No AFB in 100 OIF	NIL	Negative
1-9 AFB/100 OIF	Scanty	Positive, report actual number (e.g., AFB seen: 4/100 fields)
10-99 AFB/100 OIF	1+	Positive
1-10 AFB/field in at least 50 OIF	2+	Positive
>10 AFB/field in at least 20 OIF	3+	Positive

AFB: Acid–fast bacilli, OIF: Oil immersion fields

VARIOUS AF STRUCTURES

MODIFICATIONS OF ZN STAINING

Standard ZN stain (hot method) is the gold standard method and is used for the observation of highly AF bacteria such as M. tuberculosis, and in this staining method, 1% carbol fuchsin with heating is used as a primary stain, followed by decolorization with 25% H₂SO₄ or 3% acid-alcohol, and finally counterstained by methylene blue. To view the less AF structures/cells, this standard ZN staining had been modified. The various modified methods are as follows:^[3,8,9]

AI IN ZN STAIN

ZN staining remains a cornerstone in the diagnosis of AFB, especially M. tuberculosis and other AF infections. The stain reveals red bacilli against a blue background. Fluorescent methods like auramine staining, which show golden bacilli on a dark background, offer higher sensitivity but are costlier and more complex to use. A major challenge is the microscopic detection of these very small bacilli (2–4 μm long, 0.2–0.5 μm wide), especially in large tissue sections. Exhaustive examination is time-consuming and can lead to fatigue and reduced diagnostic accuracy. Therefore, microbiologists and pathologists usually focus on high-yield areas, such as necrotic zones and granulomas. To reduce the time of examination and chances of error, attempts at automatic detection of mycobacteria represent the logical answer to this problem.^[12]

Computer vision and machine learning (i.e., AI) have improved how we detect diseases like TB using microscope slides. These methods can automatically capture and process images, find bacteria, and use models like neural networks to identify them. Deep learning, a type of machine learning, works like the human brain and has helped make big progress since 2010, especially in reading medical images. Old glass slides are now scanned and viewed on computers at very high magnification, making it easier and faster to examine the samples. This digital method improves accuracy and helps microbiologists and pathologists work together better. Traditional ways of checking slides by hand can miss up to half of TB cases and take a long time up to hours per slide. In comparison, machines can scan more slides quickly and detect TB more reliably, reducing mistakes and saving time.^[13]

The first method for AI detection of AFB was developed on smears stained with auramine by Veropoulos et al. in 1999.^[14] Various studies have been done to evaluate the use of AI in ZN smear. Witarto et al.^[15] evaluated an AI-based algorithm for automated detection of M. tuberculosis in ZN-stained sputum smears as a potential screening tool for active TB. Developed for tissue analysis, the algorithm demonstrated high accuracy (98.33%), specificity (100%), and sensitivity (95.65%) in tissue samples. When tested on 1,059 sputum smears, it maintained 100% sensitivity and over 95% accuracy for high-confidence patches (>80%). However, specificity was modest (86.84%) due to smear variability and artifacts (e.g., dust, uneven staining). The absence of false negatives suggests that the algorithm is reliable for TB screening. Further training on sputum samples is planned to enhance specificity.^[15] Zurac et al. ^[12] also conducted a study to develop and validate a novel AI-based system for detecting M. tuberculosis in ZN-stained tissue slides. They created a training dataset of over 260,000 positive and 700 million negative image patches from 510 whole slide images. The model utilized custom computer vision architectures and image augmentation for robust learning. It generated heat maps highlighting areas likely to contain bacilli, aiding the microbiologists and pathologists in diagnosis. Internal validation and clinical testing on 60 slides demonstrated high performance: 98.33% accuracy, 95.65% sensitivity, and 100% specificity. This AI-assisted method outperformed previous approaches and reduced diagnostic workload without compromising accuracy.^[12]

Gupta et al. ^[16] conducted a prospective, multicentric clinical trial to assess the efficacy of an AI-based microscopy system in detecting AFB in ZN-stained sputum smears. The study was performed across three North Indian hospitals, with 400 samples of sputum from patients suspected of pulmonary TB. Each slide was evaluated both by expert microscopists and the AI system. The AI-based method demonstrated a sensitivity of 89.25%, specificity of 92.15%, positive predictive value of 75.45%, negative predictive value of 96.94%, and overall diagnostic accuracy of 91.53%. These findings support the utility of AI as a reliable screening tool, particularly in settings where trained microscopists are unavailable, offering a practical solution for early TB diagnosis in resource-limited regions [Table 5].^[16,17]

DISCUSSION

The ZN staining technique remains a cornerstone in microbiological diagnostics, especially for detecting M. tuberclosisand other AF structures. Over the years, various modifications of the standard ZN method have been developed to enhance its applicability across a wide range of clinical samples. Alternative techniques such as auramine fluorochrome staining provide increased sensitivity, though often at a higher cost and complexity. As diagnostic microbiology continues to evolve, the staining technique is poised for further innovation, especially at the intersection of digital pathology and automation. Future developments are likely to focus on enhancing the standardization and scalability of ZN staining through automated slide preparation and imaging systems, minimizing operator-dependent variability, and improving throughput in high-volume laboratories. The integration of AI into AFB detection represents a transformative step forward. AI-based tools now offer high accuracy, sensitivity, and specificity in screening ZN-stained smears, reducing diagnostic time and error. These AI and machine learning algorithms into routine diagnostics will continue to expand, potentially enabling real-time, cloud-based analysis of ZN-stained smears. With advances in mobile microscopy and telepathology, point-of-care AFB detection may soon become a practical reality, bridging the diagnostic gap in areas lacking skilled personnel or infrastructure. The future of ZN staining, therefore, lies not only in technical refinement but also in strategic integration with digital health platforms and national TB control programs, ensuring that even century-old methods can thrive in the era of precision diagnostics.

CONCLUSION

The Ziehl–Neelsen staining method continues to be a fundamental tool for detecting acid-fast organisms. Ongoing refinements and the integration of digital microscopy and AI are enhancing its accuracy, consistency, and applicability, ensuring its relevance in modern microbiological diagnostics. The AI algorithm demonstrated performance that was equal to or better than that of pathologists and microbiologists with varying expertise, reducing human error from fatigue and improving efficiency by cutting examination time by at least one-third. Future work will involve annotating additional positive whole side imaging for retraining, expanding clinical testing across multiple hospitals, and further refining the model to enhance robustness against diverse ZN staining techniques. Continued research and building larger datasets will help strengthen its diagnostic accuracy and clinical applicability in future.