John P. Cole, MD
Jean Marc Devroye, M
"We work basically only with donor hair."1
Surprisingly, it has taken us considerable time to place emphasis on this portion of the procedure. Conversely, two decades ago the foundation for soft natural hairlines began with the employment of single hair grafts for the hairline. This development was followed by the introduction of large numbers of natural follicular groups to the single hairs in an irregular manner to fashion a natural appearance to the recipient area. We have spent considerable effort assessing and improving the aesthetics of the hair transplant. More recently factors influencing survival and donor harvesting techniques have surfaced as areas of significant interest to hair restoration surgeons.2 Understanding these factors and assessing the yield of donor harvesting technique requires an appreciation for the contents of the donor region. The practioner must also possess an efficient means of recording the results of his dissection. He may then compare his results to the known quantities in the donor region to access the efficiency of his dissection.
There three densities, which should be evaluated with each procedure. The first is the hair density or the number of hairs per square centimeter. The second is the follicular density, which is the number of follicular groups (FG) per square centimeters. The third is the number of hairs per follicular group, which is a calculated value. Because it is calculated, I call this the calculated density. The calculated density determines the percentage of FG containing a specific number of hairs. Therefore, once we know the calculated density, we can predict the proportion of FG of any particular size. For instance, for any particular calculated density you can estimate how many three hair grafts, how many two hair grafts, and how many natural one hair grafts should result from the donor dissection. How do we know this?
First, lets discuss the follicular group. What is it? Hairs exit the skin in clusters. Some of the hairs exit from the same follicular canal, while others exit from separate but closely neighboring canals. Many in our field call these clusters follicular units. This is unfortunate because the term follicular unit is a bastard term born out of wedlock and has little more than historical significance to the average hair transplant surgeon. The follicular unit is a histological term first described by Headington in 19843. In Headington's paper he stated that the follicular unit is the pilosebaceous unit as disclosed at the mid-dermal level. Headington noted that the density of the follicular unit was about one per square millimeter. He pointed out that he did not study regional or racial variation in this density. Headington stated nothing about the surface anatomy of hair. It is the surface anatomy that we deal with in hair restoration surgery.
There is a wide variation in the follicular density from one region of the scalp to another and from one individual to another. Furthermore, there is a wide variation in the calculated density from one patient to another and from one region of the scalp to another. Individuals with a high follicular density have a reduction in the distance between the follicular groups, which results from packing more groups in the same surface area. As the distance between the groups decreases it becomes more difficult to assess whether a cluster of hair is composed of one follicular unit or more than one follicular unit (fig 1a and 1b). Another problem in assessment arises with the high calculated density. Here it becomes difficult to determine whether the large follicular group is composed of one or more than one follicular unit. In the situation where there is a high follicular density and a high calculated density, the problem is even more marked. For these reasons, it is completely inappropriate to call our natural clusters of hair bearing grafts follicular units. Rather, they should be termed follicular groups, follicular clusters, or bundle grafts.
Fig 1a. 5 hair follicular group.
Follicular groups exist in symmetrical patterns, although there is a somewhat irregular arrangement of the follicular groups within the pattern. Jimenez and Ruifernandez first noted the formula L = k / sq rt of n, where L is the density of hairs in square millimeters, k is a constant depending on the geometric spacing of the follicular units, and n is the density of hair in square centimeters. I have evaluated their formula and found it to be accurate. The geometric arrangement of follicular units follows a triangular pattern (fig 2). k is 10.7 for a triangular pattern. In this case the density of follicular units must equal 114.5 per square centimeters for the distance between the follicular units to equal 1 mm. The distance between the follicular units should be measured from the center of the follicular unit..
Jimenez and Ruifernandez are to be commended for the introduction of this powerful formula. With this formula we can make many predictions and question old beliefs. Last year I presented a paper on the regional variation of hair density and diameter. In this paper I noted that the 4 mm punch graft is elliptical in shape, not circular. For this reason the surface area is greater than that of the 4 mm circle. The variation in this surface area is noted in fig. 3.
Fig. 4. Four mm plug biopsies presented by Whiting at the 1999 New York Hair Meeting
You will first note that the surface area of the 4 mm circle is 12.57 square millimeters. You can see from this data that the mean surface area of the 4mm punch graft is greater than 12.57 mm2 in all four regions. The reason for this discrepancy results from the angle of hair growth. You will recall that the length of an incision of a two bladed knife is equal to the width between the blades divided by the sin of the angle of insertion. This fact makes the length of the 4mm plug greater than the width, an occurrence that results in an elliptical surface. Recall that Headington and Whiting performed their research by evaluating plugs, as well4. Neither took into account these geometric principals. Unfortunately, this makes their data inaccurate. Analysis of Whiting's data shows that he found an average of 13 follicular units in the 4 mm plug (fig 4.). This corresponds to a density of 1.03 FU per square millimeter. Using the Principal of Jimenez and Ruifernandez (PJR)we calculate the distance between follicular units at 1.05 mm or ½ mm greater than purported by Headington, Whiting, and Bernstein5. If you take into account that the average 4mm plug resulted in a surface area greater than 12.57 mm2, the density of follicular units is actually less than 1 per mm2. Extrapolating my mean surface area for the crown to Whiting's findings would result in an average density of 0.87 FU/ mm2. Pluging this data into PJR results in a distance of 1.15 mm between follicular units or 0.15 mm greater than the distance of Headington, Whiting, and Bernstein. This suggests that the density of follicular units and hair is less at the mid-dermal level than at the surface of the skin. It implies that our skin structure is
capable of creating more density on the surface of the skin than actually exists in the human body.
This also suggests we can define two distances between the follicular units. LSA stands for the distance between the follicular units on the surface area and LMD represents the distance between the follicular units at the mid-dermis. We see that as the density of follicular units decreases the distance between the follicular units increases and that as the density of follicular units increases, the distance between the follicular units decreases. Similary, if the distance between follicular units increases (LMD), the density of follicular units must decrease if k remains constant as compared to LSA where the distance between follicular units decreases as the density of follicular units increases. In other words, the skin structure appears generate more density from less. This principal is supported by my findings from the study of regional variation of density (fig 5.).
Fig 5. Surface Density of Follicular Groups per Square Millimeter
From this chart we see that the corrected mean surface density of follicular groups (number of follicular groups / the measured surface area) is 110 FG per cm2 in the crown. This is certainly greater than the extrapolated value of Whiting's crown density of follicular units at the mid-dermal level (87 FU per cm2.). The mean LSA of the crown is calculated at 1.02mm between follicular groups.
Ron Shapiro and Walter Unger among others have stated they are able to create the illusion of more fullness when they incorporate grafts containing more than one follicular unit. Grafts containing more than one follicular unit have a decreased LSA and consequently, a higher density than the preoperative natural density of the donor region. This powerful formula of Jiminez and Ruifernandez may shed some mathematical support to their assertions. Could it be that decreasing L improves the illusion of coverage? Indeed, there may be greater value to the larger graft than is possible from the use of pure follicular groups alone. It may be that as hair transplant surgeons, we are able to create "more from less" by incorporating a combination of grafts.
The calculated density may be assessed in different ways. First, you can preoperatively count the number of follicular groups and hairs in a specific surface area. You should exercise care to use adequate lighting and magnification during this process. I have found the Rassman Densitometer, also known as the 30X illuminated microscope (cat. No 63-851) at Radio Shack to note the follicular and hair densities and subsequently, calculate the CD. Dr. Devorye pointed out that better lighting and higher magnification improve the physician's ability to record hair counts. Therefore, the second way to obtain the CD is to count the number of hairs you see in a series of grafts and then divide the total number of hairs by the number of follicular groups you assessed. The third way to obtain the CD is to count the total number of hairs produced by the surgery staff and divide this by the number of grafts created.
In 1996 I began collecting data from graft dissection. In all 107,000 grafts were evaluated. For the purpose of this evaluation I asked my surgery staff to maintain the integrity of each follicular group and to accurately as possible record the total number of hairs in each graft produced in this manner. The dissection included the use of the Meji EMT microscope and 5X loops. I first calculated the density for each patient by dividing the total number of hairs produced by the total number of grafts. I then quantified the number of 1, 2, 3, 4, 5, and 6 hair grafts produced for that patient. I then placed all patients with the same calculated density in a single group. I then determined the mean ratio of one, two, three, four, five, and six hair follicular groups as a percentage of 100. I found that the post operative calculated densities could be applied to the pre-operative calculated density to achieve a reasonable prediction of the expected number of a particular size graft. (Fig 6 and Fig 7). The average calculated density is about 2.3 hairs per mm2.
Fig 6. Comparison of the ratio of different size follicular units as a function of the calculated density.
Fig 7. Table of Calculated densities and the corresponding percentage of each size follicular group.
<hrdata-mce-alt="Page 3" class="system-pagebreak" title="Page 3" />
An advantage of noting the calculated density of a particular technician is that it allows a rapid assessment of their efficiency. If a technician's dissection results in a lower than expected calculated density, the technician has fractionated the natural follicular groups or had a higher than expected waste. If the technician produces a higher than expected calculated density, they have produced grafts larger than natural follicular groups.
It remains to increase the sample size for the extremes of calculated densities (very high and very low calculated densities) in this study. We should also begin to look at the pre-operative calculated density and determine the ratio of one, two, 3, 4, 5, and 6 hair follicular groups prior to dissection. This sample should be evaluated carefully with a microscope. The hairs in each follicular group should be inspected manually with a pair of fine jewelers forceps to determine the actual number of hairs in each group. The results of this inspection should be compared against the aforementioned findings to assess their validity.
Last year I explained to Dr. Devroye an idea for assisting in the counting of grafts. The idea was to create a computer program that would record the number of grafts, the rate of production, the total number of hairs, the calculated density for each technician, and summarize the data. The results were to be compared to the predicted results based on a series of calculations I previously devised and the projected ratio of graft sizes based on the pre-operative calculated density. Using this data we could measure the output of hair, the efficiency of our staff, the speed of our individual staff members, and continue in our quest to maximize donor yield. I also desired to link the program to a voice recognition software package so that the technicians would have their hands freed to produce grafts. The act of writing the graft count takes away from productivity. Accordingly, it is our hope to incorporate a voice recognition program with this software to speed the graft production phase, improve the accuracy of the hair count and graft count. In this scenario a lavaliere microphone will be attached to the collar of each graft cutter. As they cut the grafts they will record the hair count of each graft. The Excel program will tabulate the total number of hairs and total graft count in a worksheet. At the end of the dissection the results will be printed and compared to the other members of the dissection team. It will be possible to keep a running record of the technician's speed of production and efficiency of dissection based on comparison with both other technicians and expected results.
Dr. Jean Devroye cleverly introduced the a power point program designed to monitor the results of the dissection and to assist the tabulation of the grafts, number of hairs, and the ratio of various size grafts. The following is his description of this software.
When I discovered the world of the hair transplantation, I was very interested in searching the accuracy of the counting of grafts and hair. Dr J.Cole gave me the opportunity to develop this subject and I would like to thank him.
Beginning with the existing database, I built an excell programming to be able to calculate the number of grafts and hair before and after cutting. The programming doesn't only involve the estimations but gives the opportunity to compare the results of the cutting obtained with the theoretical expectations.
This includes the work of all the team or specifically of each technician.
For the ones who were in Hawaii, you can notice the modifications because the study is in progress and any comment is welcome.
Counting before cutting
Compared to the old one, the new version is easier to use. In order to use the programming, the first step is to measure the strip surface and the hair density. I took the usual repartition of the different types of FU according to the hair density. (based on Dr Cole's chart).
For example, if you base your FU density on 10 FU per 10 mm square, and you observe a hair density equal of 16 hair per 10 mm square, you will find 4 singles and 6 doubles.
Mathematics confirm the theory: 4*1 + 6*2 = 16 hair.
Each hair density usually follows the repartition of the graph above. But we are working on refining the database in function of different criteria (ethnical origins, etc...).
Example: After measurement, you will obtain this board.
|Number of FU||380||967||527||127||
Counting after cutting
The second part of this programming is to study the results after the cutting.
Again, it's very interesting to compare the theoretical goals with the results obtained.
Using another board, we introduce the results of the cutting (either the ones of each technician, or the ones of all the team together)
We also built a graphic pages which gives us a general view of the results.
Now, we have the project to use the speech recognition system.
Such a performing system will help hair transplant surgery to work fast and precise.
Arnold, James: Cyberspace Chat 7/4/2001
Cey, Victoria: 2001 ISHRS Opinion poll
Headington JT. Transverse microscopic anatomy of the human scalp. Arch Dermatol 1984;120:449-56.
Whiting D: New York Hair Meeting 1999
Bernstein RM, Rassman WR:Follicular Transplantation Patient Evaluation and Surgical Planning. Dermatologic Surgery 23:771-784, 1997 and Bernstein R, Rassman W, Szaniawski W, Halperin A: Follicular Transplantation. International Journal of Aesthetic and Restorative Surgery 3:119-132, 1995.